Kenneth Downs

Today's essay deals with the tricky issue of custom features for individual customers who are running instances of your software.

The question comes by way of a regular reader who prefers to remain anonymous, but asks this:

... I work on a large (to me, anyway) application that serves as a client database, ticket system, time-tracking, billing, asset-tracking system. We have some customers using their own instances of the software. Often, those customers want additional fields put in different places (e.g., a priority column on tickets). This results in having multiple branches to account for versions with slight changes in code and in the database. This makes things painful and time-consuming in the long run: applying commits from master to the other branches requires testing on every branch; same with database migrate scripts, which frequently have to be modified.

Is there an easier way? I have thought about the possibility of making things "optional" in the database, such as a column on a table, and hiding its existence in the code when it's not "enabled." This would have the benefit of a single code set and a single database schema, but I think it might lead to more dependence on the code and less on the database -- for example, it might mean constraints and keys couldn't be used in certain cases.

Our reader asks, is it better to have different code branches or to try to keep a lot of potentially conflicting and optional items mixed in together?

Well, the wisdom of the ages is to maintain a single code branch, including the database schema. I tried exactly once, very early in my career, to fork my own code, and gave up almost within days. When I went to work in larger shops I always arrived in a situation where the decision had already been made to maintain a single branch. Funny thing, since most programmers cannot agree on the color of the sky when they're staring out the window, this is the only decision I have ever seen maintained with absolute unanimity no matter how many difficulties came out of it.

There is some simple arithmetic as to why this is so. If you have single feature for a customer that is giving you a headache, and you fork the code, you now have to update both code branches for every change plus regression test them both, including the feature that caused the headache. But if you keep them combined you only have the one headache feature to deal with. That's why people keep them together.

Making custom features work smoothly is a two-step process. The first step is arguably more difficult than the second, but the second step is absolutely crucial if you have business logic tied to the feature.

Most programmers when confronted with this situation will attempt to make various features optional. I consider this to be a mistake because it complicates code, especially when we get to step 2. By far the better solution is to make features ignorable by anybody who does not want them.

The wonderful thing about ingorable features is they tend to eliminate the problems with apparently conflicting features. If you can rig the features so anybody can use either or both, you've eliminated the conflict.

As mentioned above, the first step is arguably more difficult than the second, because it may involve casting requirements differently than they are presented.

For example, our reader asks about a priority column on tickets, asked for by only one customer. This may seem like a conflict because nobody else wants it, but we can dissolve the conflict when we make the feature ignorable. The first step involves doing this at the database or schema level.

But first we should mention that the UI is easy, we might have a control panel where we can make fields invisible. Or maybe our users just ignore the fields they are not interested in. Either way works.

The problem is in the database. If the values for priority come from a lookup table, which they should, then we have a foreign key, and we have a problem if we try to ignore it:

• We can allow nulls in the foreign key, which is fine for the people ignoring it, but
• This means the people who require it can end up with tickets that have no priority because it does not prevent a user from leaving it blank.

A simple answer here is to pre-populate your priority lookup table with a value of "Not applicable", perhaps with a hardcoded id of zero. Then we set the default value for the TICKET.priority to zero. This means people can safely ignore it because it will always be valid.

Then, for the customer who paid for it, we just go in after the install and delete the default entry. It's a one-time operation, not even worth writing a script for, and it forces them to create a set of priorities before using the system. Further, by leaving the default of zero in there, it forces valid answers because users will be dinged with an FK violation if they do not provide a real priority.

For this particular example, there is no step 2, because the problem is completely solved at the schema level. To see how to work with step 2, I will make up an example of my own.

To illustrate step 2, I'm going to make up an example that is not really appropriate to our reader's question, frankly because I cannot think of one for that situation.

Let's say we have an eCommerce system, and one of our sites wants customer-level discounts based on customer groups, while another wants discounts based on volume of order -- the more you buy, the deeper the discount. At this point most programmers start shouting in the meeting, "We'll make them optional!" Big mistake, because it makes for lots of work. Instead we will make them ignorable.

Step 1 is to make ignorable features in the schema. Our common code base contains a table of customer groups with a discount percent, and in the customers table we make a nullable foreign key to the customer groups table. If anybody wants to use it, great, and if they want to ignore it, that's also fine. We do the same thing with a table of discount amounts, we make an empty table that lists threshhold amounts and discount percents. If anybody wants to use it they fill it in, everybody else leaves it blank.

Now for the business logic, the calculations of these two discounts. The crucial idea here is not to make up conditional logic that tries to figure out whether or not to apply the discounts. It is vastly easier to always apply both discounts, with the discounts coming out zero for those users who have ignored the features.

So for the customer discount, if the customer's entry for customer group is null, it will not match to any discount, and you treat this as zero. Same for the sale amount discount, the lookup to see which sale amount they qualify doesn't find anything because the table is empty, so it treats it as zero.

So the real trick at the business logic level is not to figure out which feature to use, which leads to complicatec conditionals that always end up conflicting with each other, but to always use all features and code them so they have no effect when they are being ignored.

Once upon a time almost everybody coding for a living dealt with these situations -- we all wrote code that was going to ship off to live at our customer's site. Nowadays this is less common, but for those of us dealing with it it is a big deal.

The wisdom of the ages is to maintain a common code base. The method suggested here takes that idea to its most complete implementation, a totally common code base in which all features are active all of the time, with no conditionals or optional features (except perhaps in the UI and on printed reports), and with schema and business logic set up so that features that are being ignored simply have no effect on the user.

The past three posts of this little mini-series have gone from a Working definition of business logic to a Rigorous definition of business logic and on to some theorems about business logic. To wrap things up, I'd like to ask the question, is it possible to isolate business logic into a single tier?

There are plenty of opinions out there. For a pretty thorough explanation of how to put everything into the DBMS, check out Toon Koppelaar's description. Mr. Koppelaars has some good material, but you do need to read through his earlier posts to get the definitions of some of his terms. You can also follow his links through to some high quality discussions elsewhere.

Contrasting Mr. Koppelaar's opinion is a piece which does not have nearly the same impact, IMHO, because in Dude, Where's My Business Logic? we get some solid history mixed with normative assertions based on either anecdote or nothing at all. I'm a big believer in anecdote, but when I read a sentence that says, "The database should not have any knowledge of what a customer is, but only of the elements that are used to store a customer." then I figure I'm dealing with somebody who needs to see a bit more of the world.

First, let's review that our rigorous definition of business logic includes schema (types and constraints), derived values (timestamps, userstamps, calculations, histories), non-algorithmic compound operations (like batch billing) and algorithmic compound operations, those that require looping in their code. This encompasses everything we might do from the simplest passive things like a constraint that prevents discounts from being over 100% to the most complex hours-long business process, along with everything in between accounted for.

Now I want to start out by using that definition to see a little bit about what is going on in the User Interface. This is not the presentation layer as it is often called but the interaction layer and even the command layer.

Consider an admin interface to a database, where the user is entering or modifying prices for the price list. Now, if the user could enter "Kim Stanley Robinson" as the price, that would be kind of silly, so of course the numeric inputs only allow numeric values. Same goes for dates.

So the foundation of usability for a UI is at very least knowlege of and enforcement of types in the UI layer. Don't be scared off that I am claiming the UI is enforcing anything, we'll get to that a little lower down.

Now consider the case where the user is typing in a discount rate for this or that, and a discount is not allowed to be over 100%. The UI really ought to enforce this, otherwise the user's time is wasted when she enters an invalid value, finishes the entire form, and only then gets an error when she tries to save. In the database world we call this a constraint, so the UI needs to know about constraints to better serve the user.

Now this same user is typing a form where there is an entry for US State. The allowed values are in a table in the database, and it would be nice if the user had a drop-down list, and one that was auto-suggesting as the user typed. Of course the easiest way to do something like this is just make sure the UI form "knows" that this field is a foreign key to the STATES table, so it can generate the list using some generic library function that grabs a couple of columns out of the STATES table. Of course, this kind of lookup thing will be happening all over the place, so it would work well if the UI knew about and enforced foreign keys during entry.

And I suppose the user might at some point be entering a purchase order. The purchase order is automatically stamped with today's date. The user might see it, but not be able to change it, so now our UI knows about system-generated values.

Is this user allowed to delete a customer? If not, the button should either be grayed out or not be there at all. The UI needs to know about and enforce some security.

So in fact the UI layer not only knows the logic but is enforcing it. It is enforcing it for two reasons, to improve the user experience with date pickers, lists, and so forth, and to prevent the user from entering invalid data and wasting round trips.

And yet, because we cannot trust what comes in to the web server over the wire, we have to enforce every single rule a second time when we commit the data.

You usually do not hear people say that the UI enforces business logic. They usually say the opposite. But the UI does enforce business logic. The problem is, everything the UI enforces has to be enforced again. That may be why we often overlook the fact that it is doing so.

Now let's go through the stuff the UI is enforcing, and see what happens in the application and the database.

With respect to type, a strongly typed language will throw an error if the type is wrong, and a weakly typed language is wise to put in a type check anyway. The the DBMS is going to only allow correctly typed values, so, including the UI, type is enforced three times.

With respect to lookups like US state, in a SQL database we always let the server do that with a foreign key, if we know what is good for us. That makes double enforcement for lookups.

So we can see where this is going. As we look at constraints and security and anything else that must be right, we find it will be enforced at least twice, and as much as three times.

By defining First Order Business Logic, the simplest foundation layer, as including things like types and keys and constraints, we find that the enforcement of this First Order stuff is done 2 or 3 times, but never only once.

This more or less leaves in tatters the idea of a "Business Logic Layer" that is in any way capable of handling all business logic all by its lonesome. The UI layer is completely useless unless it is also enforcing as much logic as possible, and even when we leave the Database Server as the final enforcer of First Order Business Logic (types, constraints, keys), it is still often good engineering to do some checks to prevent expensive wasted trips to the server.

So we are wasting time if we sit around trying to figure out how to get the Business Logic "where it belongs", because it "belongs" in at least two places and sometimes three. Herding the cats into a single pen is a fool's errand, it is at once unnecessary, undesirable, and impossible.

Update: Regular reader Dean Thrasher of Infovark summarizes most of what I'm saying here using an apt industry standard term: Business Logic is a cross-cutting concern.

Only when we have smashed the concept that Business Logic can exist in serene isolation in its own layer can we start to ask the questions that would actually speed up development and make for better engineering.

Freed of the illusion of a separate layer, when we look at the higher Third and Fourth Order Business Logic, which always require coding, we can decide where they go based either on engineering or the availability of qualified programmers in particular technologies, but we should not make the mistake of believing they are going where they go because the gods would have it so.

But the real pressing question if we are seeking to create efficient manageable large systems is this: how we distribute the same business logic into 2 or 3 (or more) different places so that it is enforced consistently everywhere. Because a smaller code base is always easier to manage than a large one, and because configuration is always easier than coding, this comes down to meta-data, or if you prefer, a data dictionary. That's the trick that always worked for me.

Anybody who disagrees with the thesis here has only to say, "Ken, those things are not business logic just because you wrote a blog that says they are. In my world business logic is about code baby!" Well sure, have it your way. After all, the nice thing about definitions is that we can all pick the ones we like.

But these definitions, the theorems I derived on Tuesday, and the multiple-enforcement thesis presented here today should make sense to anbyody struggling with where to put the business logic. That struggle and its frustrations come from the mistake of imposing abstract conceptual responsibilities on each tier instead of using the tiers as each is able to get the job done. Databases are wonderful for type, entity integrity (uniqueness), referential integrity, ACID compliance, and many other things. Use them! Code is often better when the problem at hand cannot be solved with a combination of keys and constraints (Fourth Order Business Logic), but even that code can be put into the DB or in the application.

So beware of paradigms that assign responsibility without compromise to this or that tier. It cannot be done. Don't be afraid to use code for doing things that require structured imperative step-wise operations, and don't be afraid to use the database for what it is good for, and leave the arguments about "where everything belongs" to those with too much time on their hands.

In yesterday's Rigorous Definition of Business Logic, we saw that business logic can be defined in four orders:

• First Order Business Logic is entities and attributes that users (or other agents) can save, and the security rules that govern read/write access to the entitites and attributes.
• Second Order Business Logic is entities and attributes derived by rules and formulas, such as calculated values and history tables.
• Third Order Business Logic are non-algorithmic compound operations (no structure or looping is required in expressing the solution), such as a month-end batch billing or, for the old-timers out there, a year-end general ledger roll-up.
• Fourth Order Business Logic are algorithmic compound operations. These occur when the action of one step affects the input to future steps. One example is ERP Allocation.

The best way to see if these have any value is to cook up some theorems and examine them with an example. We will take a vastly simplified time billing system, in which employees enter time which is billed once/month to customers. We'll work out some details a little below.

The first theorem we can derive from these definitions is that we should look at First and Second Order Schemas together during analysis. This is because:

• First Order Business Logic is about entities and atrributes
• Second Order Business Logic is about entities and attributes
• Second Order Business Logic is about values generated from First Order values and, possibly, other Second Order values
• Therefore, Second Order values are always expressed ultimately in terms of First Order values
• Therefore, they should be analyzed together

To give the devil his due, ORM does this easily, because it ignores so much database theory (paying a large price in performance for doing so) and considers an entire row, with its first order and second order values together, as being part of one class. This is likely the foundation for the claims of ORM users that they experience productivity gains when using ORM. Since I usually do nothing but bash ORM, I hope this statement will be taken as utterly sincere.

Going the other way, database theorists and evangelists who adhere to full normalization can hobble an analysis effort by refusing to consider 2nd order because those values denormalize the database, so sometimes the worst of my own crowd will prevent analysis by trying to keep these out of the conversation. So, assuming I have not pissed off my own friends, let's keep going.

So let's look at our case study of the time billing system. By theorem 1, our analysis of entities and attributes should include both 1st and 2nd order schema, something like this:

 
 INVOICES
-----------
 invoiceid      2nd Order, a generated unique value
 date           2nd Order if always takes date of batch run
 customer       2nd Order, a consequence of this being an
                           aggregation of INVOICE_LINES
 total_amount   2nd Order, a sum from INVOICE_LINES
               
 INVOICE_LINES
---------------
 invoiceid      2nd order, copied from INVOICES
 customer         +-  All three are 2nd order, a consequence
 employee         |   of this being an aggregration of
 activity         +-  employee time entries
 rate           2nd order, taken from ACTIVITIES table
                           (not depicted)
 hours          2nd order, summed from time entries
 amount         2nd order, rate * hours
 
 TIME_ENTRIES
--------------
 employeeid     2nd order, assuming system forces this
                    value to be the employee making
                    the entry
 date           1st order, entered by employee
 customer       1st order, entered by employee
 activity       1st order, entered by employee
 hours          1st order, entered by employee

Now, considering how much of that is 2nd order, which is almost all of it, the theorem is not only supported by the definition, but ought to line up squarely with our experience. Who would want to try to analyze this and claim that all the 2nd order stuff should not be there?

The second theorem we can derive from these definitions is that First and Second Order Business logic require separate implementation techniques. This is because:

• First Order Business Logic is about user-supplied values
• Second Order Business Logic is about generated values
• Therefore, unlike things cannot be implemented with like tools.

Going back to the time entry example, let's zoom in on the lowest table, the TIME_ENTRIES. The employee entering her time must supply customer, date, activity, and hours, while the system forces the value of employeeid. This means that customer and activity must be validated in their respective tables, and hours must be checked for something like <= 24. But for employeeid the system provides the value out of its context. So the two kinds of values are processed in very unlike ways. It seems reasonable that our code would be simpler if it did not try to force both kinds of values down the same validation pipe.

This theorem states that the sum of Second and Third Order Business Logic is fixed:

• Second Order Business Logic is about generating entities and attributes by rules or formulas
• Third Order Business Logic is coded compound creation of entities and attributes
• Given that a particular set of requirements resolves to a finite set of actions that generate entities and values, then
• The sum of Second Order and Third Order Business Logic is fixed.

In plain English, this means that the more Business Logic you can implement through 2nd Order declarative rules and formulas, the fewer processing routines you have to code. Or, if you prefer, the more processes you code, the fewer declarative rules about entitities and attributes you will have.

This theorem may be hard to compare to experience for verification because most of us are so used to thinking in terms of the batch billing as a process that we cannot imagine it being implemented any other way: how exactly am I suppose to implement batch billing declaratively?.

Let's go back to the schema above, where we can realize upon examination that the entirety of the batch billing "process" has been detailed in a 2nd Order Schema, if we could somehow add these facts to our CREATE TABLE commands the way we add keys, types, and constraints, batch billing would occur without the batch part.

Consider this. Imagine that a user enters a a TIME_ENTRY. The system checks for a matching EMPLOYEE/CUSTOMER/ACTIVITY row in INVOICE_DETAIL, and when it finds the row it updates the totals. But if it does not find one then it creates one! Creation of the INVOICE_DETAIL record causes the system to check for the existence of an invoice for that customer, and when it does not find one it creates it and initializes the totals. Subsequent time entries not only update the INVOICE_DETAIL rows but the INVOICE rows as well. If this were happening, there would be no batch billing at the end of the month because the invoices would all be sitting there ready to go when the last time entry was made.

By the way, I coded something that does this in a pretty straight-forward way a few years ago, meaning you could skip the batch billing process and add a few details to a schema that would cause the database to behave exactly as described above. Although the the format for specifying these extra features was easy enough (so it seemed to me as the author), it seemed the conceptual shift of thinking that it required of people was far larger than I initially and naively imagined. Nevertheless, I toil forward, and that is the core idea behind my Triangulum project. Observation: There Will Be Code

This is not so much a theorem as an observation. This observation is that if your application requires Fourth Order Business Logic then somebody is going to code something somewhere.

An anonymous reader pointed out in the comments to Part 2 that Oracle's MODEL clause may work in some cases. I would assume so, but I would also assume that reality can create complicated Fourth Order cases faster than SQL can evolve. Maybe.

But anyway, the real observation here is is that no modern language, either app level or SQL flavor, can express an algorithm declaratively. In other words, no combination of keys, constraints, calculations and derivations, and no known combination of advanced SQL functions and clauses will express an ERP Allocation routine or a Magazine Regulation routine. So you have to code it. This may not always be true, but I think it is true now.

This is in contrast to the example given in the previous section about the fixed total of 2nd and 3rd Order Logic. Unlike that example, you cannot provide enough 2nd order wizardry to eliminate fourth order. (well ok maybe you can, but I haven't figured it out yet myself and have never heard that anybody else is even trying. The trick would be to have a table that you truncate and insert a single row into, a trigger would fire that would know how to generate the next INSERT, generating a cascade. Of course, since this happens in a transaction, if you end up generating 100,000 inserts this might be a bad idea ha ha.)

This theorem rests on the acceptance of an observation, that using meta-data repositories, or data dictionaries, is easier than coding. If that does not hold true, then this theorem does not hold true. But if that observation (my own observation, admittedly) does hold true, then:

• By Theorem 3, the sum of 2nd and 3rd order logic is fixed
• By observation, using meta-data that manages schema requires less time than coding,
• By Theorem 1, 2nd order is analyzed and specified as schema
• Then it is desirable to specify as much business logic as possible as 2nd order schema, reducing and possibly eliminating manual coding of Third Order programs.

Again we go back to the batch billing example. Is it possible to convert it all to 2nd Order as described above. Well yes it is, because I've done it. The trick is an extremely counter-intuitive modification to a foreign key that causes a failure to actually generate the parent row that would let the key succeed. To find out more about this, check out Triangulum (not ready for prime time as of this writing).

The major conclusion in all of this is that anlaysis and design should begin with First and Second Order Business Logic, which means working out schemas, both the user-supplied values and the system-supplied values.

When that is done, what we often call "processes" are layered on top of this.

Tomorrow we will see part 4 of 4, examining the business logic layer, asking, is it possible to create a pure business logic layer that gathers all business logic unto itself?

This is part 2 of a 4 part mini-series that began before the holidays with A Working Definition Business Logic. Today we proceed to a rigorous definition, tomorrow we will see some theorems, and the series will wrap up with a post on the "business layer."

In the first post, the working definition said that business logic includes at least:

• The Schema
• Calculations
• Processes

None of these was very rigorously defined, kind of a "I'll know it when I see it" type of thing, and we did not talk at all about security. Now the task becomes tightening this up into a rigorous definition.

Toon Koppelaars has some excellent material along these same lines, and a good place to start is his Helsinki Declaration (IT Version). The articles have a different focus than this series, so they make great contrasting reading. I consider my time spent reading through it very well spent.

What I propose below is a definition in four parts. As definitions, they are not supposed to prove anything, but they are definitely supposed to ring true to the experience of any developer who has created or worked on a non-trivial business application. This effort would be a success if we reach some concensus that "at least it's all in there", even if we go on to argue bitterly about which components should be included in which layers.

Also, while I claim the definitions below are rigorous, they are not yet formal. My instinct is that formal definitions can be developed using First Order Logic, which would allow the theorems we will see tomorrow to move from "yeah that sounds about right" to being formally provable.

As for their practical benefit, inasmuch as "the truth shall make you free", we ought to be able to improve our architectures if we can settle at very least what we are talking about when we use the vague term "business logic."

What we commonly call "business logic", by which we vaguely mean, "That stuff I have to code up", can in fact be rigorously defined as having four parts, which I believe are best termed orders, as there is a definite precedence to their discovery, analysis and implementation.

• First Order: Schema
• Second Order: Derivations
• Third Order: Non-algorithmic compound operations
• Fourth Order: Algorithmic compound operations

Now we examine each order in detail.

Even "schema-less" databases have a schema, they simply do not enforce it in the database server. Consider: an eCommerce site using MongoDB is not going to be tracking the local zoo's animal feeding schedule, because that is out of scope. No, the code is limited to dealing with orders, order lines, customers, items and stuff like that.

It is in the very act of expressing scope as "the data values we will handle" that a schema is developed. This holds true regardless of whether the datastore will be a filesystem, an RDBMS, a new NoSQL database, or anything else.

Because all applications have a schema, whether the database server enforces it or whether the application enforces it, we need a vocabulary to discuss the schema. Here we have an embarrasment of choices, we can talk about entities and attributes, classes and properties, documents and values, or columns and tables. The choice of "entities and attributes" is likely best because it is as close as possible to an implementation-agnostic language.

We can define schema, including security, as:

that body of entities and their attributes whose relationships and values will be managed by the application stack, including the authorization of roles to read or write to entities and properties.

Schema in this definition does not include derived values of any kind or the processes that may operate on the schema values, those are higher order of business logic. This means that the schema actually defines the entire body of values that the application will accept from outside sources (users and other programs) and commit to the datastore. Restating again into even more practical terms, the schema is the stuff users can save themselves.

With all of that said, let's enumerate the properties of a schema.

Type is required for every attribute.

Constraints are limits to the values allowed for an attribute beyond its type. We may have a discount percent that may not exceed 1.0 or 100%.

Entity Integrity is usually thought of in terms of primary keys and the vague statement "you can't have duplicates." We cannot have a list of US States where "NY" is listed 4 times.

Referential Integrity means that when one entity links or refers to another entity, it must always refer to an existing entity. We cannot have some script kiddie flooding our site with sales of items "EAT_ME" and "F***_YOU", becuase those are not valid items.

The general term 'validation' is not included because any particular validation rule is is a combination of any or all of type, constraints, and integrity rules.

When we speak of derived values, we usually mean calculated values, but some derivations are not arithmetic, so the more general term "derived" is better. Derivations are:

A complete entity or an attribute of an entity generated from other entities or attributes according to a formula or rule.

The definition is sufficiently general that a "formula or rule" can include conditional logic.

Simple arithmetic derived values include things like calculating price * qty, or summing an order total.

Simple non-arithmetic derivations include things like fetching the price of an item to use on an order line. The price in the order is defined as being a copy of the item's price at the time of purchase.

An example of a complete entity being derived is a history table that tracks changes in some other table. This can also be implemented in NoSQL as a set of documents tracking the changes to some original document.

Security also applies to generated values only insofar as who can see them. But security is not an issue for writing these values because by definition they are generated from formulas and rules, and so no outside user can ever attempt to explicitly specify the value of a derived entity or property.

One final point about Second Order Business Logic is that it can be expressed declaratively, if we have the tools, which we do not, at least not in common use. I wrote one myself some years ago and am re-releasing it as Triangulum, but that is a post for another day.

The definitions of First and Second Order Business Logic have the advantage of being agnostic to what kind of datastore you are using, and being agnostic to whether or not the derived values are materialized. (In relational terms, derivations are almost always denormalizing if materialized, so in a fully normalized database they will not be there, and you have to go through the application to get them.)

Nevertheless, these two definitions can right off bring some confusion to the term "schema." Example: a history table is absolutely in a database schema, but I have called First Order Business Logic "schema" and Second Order Business Logic is, well, something else. The best solution here is to simply use the terms First Order Schema and Second Order Schema. An order_lines table is First Order schema, and the table holding its history is Second Order Schema.

The now ubiquitous auto-incremented surrogate primary keys pose another stumbling block. Because they are used so often (and so often because of seriously faulty reasoning, see A Sane Approach To Choosing Primary Keys) they would automatically be considered schema -- one of the very basic values of a sales order, check, etc. But they are system-generated so they must be Second Order, no? Isn't the orderid a very basic part of the schema and therefore First Order? No. In fact, by these definitions, very little if any of an order header is First Order, the tiny fragments that are first order might be the shipping address, the user's choice of shipping method, and payment details provided by the user. The other information that is system-generated, like Date, OrderId, and order total are all Second Order.

Before defining Third Order Business Logic I would like to offer a simple example: Batch Billing. A consulting company bills by the hour. Employees enter time tickets throughout the day. At the end of the month the billing agent runs a program that, in SQL terms:

• Inserts a row into INVOICES for each customer with any time entries
• Inserts a row into INVOICE_LINES that aggregates the time for each employee/customer combination.

This example ought to make clear what I mean by definining Third Order Business Logic as:

A Non algorithmic compound operation.

The "non-algorithmic" part comes from the fact that none of the individual documents, an INVOICE row and its INVOICE_LINES, is dependent on any other. There is no case in which the invoice for one customer will influence the value of the invoice for another. You do not need an algorithm to do the job, just one or more steps that may have to go in a certain order.

Put another way, it is a one-pass set-oriented operation. The fact that it must be executed in two steps is an artifact of how database servers deal with referential integrity, which is that you need the headers before you can put in the detail. In fact, when using a NoSQL database, it may be possible to insert the complete set of documents in one command, since the lines can be nested directly into the invoices.

Put yet a third way, in more practical terms, there is no conditional or looping logic required to specify the operation. This does not mean there will be no looping logic in the final implementation, because performance concerns and locking concerns may cause it to be implemented with 'chunking' or other strategies, but the important point is that the specification does not include loops or step-wise operations because the individual invoices are all functionally independent of each other.

I do not want to get side-tracked here, but I have had a working hypothesis in my mind for almost 7 years that Third Order Business Logic, even before I called it that, is an artifact, which appears necessary because of the limitations of our tools. In future posts I would like to show how a fully developed understanding and implementation of Second Order Business Logic can dissolve many cases of Third Order.

We now come to the upper bound of complexity for business logic, Fourth Order, which we label "algorithmic compound operations", and define a particular Fourth Order Business Logic process as:

Any operation where it is possible or certain that there will be at least two steps, X and Y, such that the result of Step X modifies the inputs available to Step Y.

In comparison to Third Order:

• In Third Order the results are independent of one another, in Fourth Order they are not.
• In Third Order no conditional or branching is required to express the solution, while in Fourth Order conditional, looping, or branching logic will be present in the expression of the solution.

Let's look at the example of ERP Allocation. In the interest of brevity, I am going to skip most of the explanation of the ERP Allocation algorithm and stick to this basic review: a company has a list of sales orders (demand) and a list of purchase orders (supply). Sales orders come in through EDI, and at least once/day the purchasing department must match supply to demand to find out what they need to order. Here is an unrealistically simple example of the supply and demand they might be facing:

  *** DEMAND ***          *** SUPPLY ***

    DATE    | QTY           DATE    | QTY
------------+-----      ------------+----- 
  3/ 1/2011 |  5          3/ 1/2011 |  3
  3/15/2011 | 15          3/ 3/2011 |  6
  4/ 1/2011 | 10          3/15/2011 | 20
  4/ 3/2011 |  7   

The desired output of the ERP Allocation might look like this:

 *** DEMAND ***      *** SUPPLY ****
    DATE    | QTY |  DATE_IN   | QTY  | FINAL 
------------+-----+------------+------+-------
  3/ 1/2011 |  5  |  3/ 1/2011 |  3   |  no
                  |  3/ 3/2011 |  2   | Yes 
  3/15/2011 | 15  |  3/ 3/2011 |  4   |  no
                  |  3/15/2011 | 11   | Yes
  4/ 1/2011 | 10  |  3/15/2011 |  9   |  no
  4/ 3/2011 |  7  |    null    | null |  no

From this the purchasing agents know that the Sales Order that ships on 3/1 will be two days late, and the Sales Orders that will ship on 4/1 and 4/3 cannot be filled completely. They have to order more stuff.

Now for the killer question: Can the desired output be generated in a single SQL query? The answer is no, not even with Common Table Expressions or other recursive constructs. The reason is that each match-up of a purchase order to a sales order modifies the supply available to the next sales order. Or, to use the definition of Fourth Order Business Logic, each iteration will consume some supply and so will affect the inputs available to the next step.

We can see this most clearly if we look at some pseudo-code:

for each sales order by date {
   while sales order demand not met {
      get earliest purchase order w/qty avial > 0
         break if none
      make entry in matching table
      // This is the write operation that 
      // means we have Fourth Order Business Logic
      reduce available qty of purchase order
   }
   break if no more purchase orders
}

As stated in the beginning, it is my belief that these four orders should "ring true" with any developer who has experience with non-trivial business applications. Though we may dispute terminology and argue over edge cases, the recognition and naming of the Four Orders should be of immediate benefit during analysis, design, coding, and refactoring. They rigorously establish both the minimum and maximum bounds of complexity while also filling in the two kinds of actions we all take between those bounds. They are datamodel agnostic, and even agnostic to implementation strategies within data models (like the normalize/denormalize debate in relational).

But their true power is in providing a framework of thought for the process of synthesizing requirements into a specification and from there an implementation.

Tomorrow we will see some theorems that we can derive from these definitions.

Update: the Second Post of this series is now available.

Update: the Third Post of this series is now available.

The Wikipedia entry on "Business Logic" has a wonderfully honest opening sentence stating that "Business logic, or domain logic, is a non-technical term... (emphasis mine)". If this is true, that the term is non-technical, or if you like, non-rigorous, then most of us spend the better part of our efforts working on something that does not even have a definition. Kind of scary.

Is it possible to come up with a decent working definition of business logic? It is certainly worth a try. This post is the first in a four part series. The second post is about a more rigorous definition of Business Logic.

This blog has two tables of contents, the Complete Table of Contents and the list of Database Skills.

In this essay we will pursue a method of finding operations that we can define as business logic with a minimum of controversey, and identify those that can likely be excluded with a minimum of controversey. This may leave a bit of gray area that can be taken up in a later post.

If we define Presentation Requirements as all requirements about "how it looks" as opposed to "what it is", then we can rule these out. But if we want to be rigorous we have to be clear, Presentation Requirements has to mean things like branding, skinning, accessibility, any and all formatting, and anything else that is about the appearance and not about the actual values fetched from somewhere.

Database veterans are likely to agree that your table schema constitutes the foundation layer of all business rules. The schema, being the tables, columns, and keys, determines what must be provided and what must be excluded. If these are not business logic, I guess I don't know what is.

What about CouchDB and MongoDB and others that do not require a predefined schema? These systems give up the advantages of a fixed schema for scalability and simplicity. I would argue here that the schema has not disappeared, it has simply moved into the code that writes documents to the database. Unless the programmer wants a nightmare of chaos and confusion, he will enforce some document structure in the code, and so I still think it safe to say that even for these databases there is a schema somewhere that governs what must be stored and what must be excluded.

So we have at least a foundation for a rigorous definition of business rules: the schema, be it enforced by the database itself or by the code, forms the bottom layer of the business logic.

The next easy addition to our definition of business logic would be processes, where a process can be defined loosely as anything involving multiple statements, can run without user interaction, may depend on parameters tables, and may take longer than a user is willing to wait, requiring background processing.

I am sure we can all agree this is business logic, but as long as we are trying to be rigorous, we might say it is business logic because:

• It must be coded
• The algorithm(s) must be inferred from the requirements
• It is entirely independent of Presentation Requirements

We also should be able to agree that calculated values like an order total, and the total after tax and freight, are business logic. These are things we must code for to take user-supplied values and complete some picture.

The reasons are the same as for processes, they must be coded, the formulas must often be inferred from requirements (or forced out of The Explainer at gunpoint), and the formulas are completely independent of Presentation Requirements.

So far we have excluded "mere" Presentation Requirements, and included three entries I hope so far are non-controversial:

• Schema
• Processes
• Calculations

These are three things that some programmer must design and code. The schema, either in a conventional relational database or in application code. Processes, which definitely must be coded, and calculations, which also have to be coded.

Plenty. At very least security and notifications. But let's put those off for another day and see how we might handle what we have so far.

For the Schema, I have already mentioned that you can either put it into a Relational database or manage it in application code when using a "NoSQL" database. More than that will have to wait for 2011, when I am hoping to run a series detailing different ways to implement schemas. I'm kind of excited to play around with CouchDB or MongoDB.

For processes, I have a separate post that examines the implications of the stored procedure route, the embedded SQL route, and the ORM route.

This leaves calculations. Let us now see how we might handle calculations.

But before we get to CRUD, I should state that if your CRUD code involves processes, seek professional help immediately. Mixing processes into CRUD is an extremely common design error, and it can be devastating. It can be recognized when somebody says, "Yes, but when the salesman closes the sale we have to pick this up and move it over there, and then we have to...."

Alas, this post is running long already and so I cannot go into exactly how to solve these, but the solution will always be one of these:

• Spawning a background job to run the process asynchronously. Easy because you don't have to recode much, but highly suspicous.
• Examining why it seems necessary to do so much work on what ought to be a single INSERT into a sales table, with perhaps a few extra rows with some details. Much the better solution, but often very hard to see without a second pair of eyes to help you out.

So now we can move on to pure CRUD operations.

Outbound CRUD is any application code that grabs data from the database and passes it up to the Presentation layer.

A fully normalized database will, in appropriate cases, require business logic of the calculations variety, otherwise the display is not complete and meaningful to the user. There is really no getting around it in those cases.

However, a database Denormalized With Calculated Values requires no business logic for outbound CRUD, it only has to pick up what is asked for and pass it up. This is the route I prefer myself.

Deciding whether or not to include denormalized calculated values has heavy implications for the architecture of your system, but before we see why, we have to look at inbound CRUD.

Inbound CRUD, in terms of business logic, is the mirror image of outbound. If your database is fully normalized, inbound CRUD should be free of business logic, since it is simply taking requests and pushing them to the database. However, if you are denormalizing by adding derived values, then it has to be done on the way in, so inbound CRUD code must contain business logic code of the calculations variety.

Now let us examine how the normalization question affects system architecture and application code.

As stated above, denormalizing by including derived values forces calculated business logic on the inbound path, but frees your outbound path to be the "fast lane". The opposite decision, not storing calculated values, allows the inbound path to be the "fast lane" and forces the calculations into the outbound path.

The important conclusion is: if you have business logic of the calculation variety in both lanes then you may have some inconsistent practices, and there may be some gain involved in sorting those out.

But the two paths are not perfectly symmetric. Even a fully normalized database will often, sooner or later, commit those calculated values to columns. This usually happens when some definition of finality is met. Therefore, since the inbound path is more likely to contain calculations in any case, the two options are not really balanced. This is one reason why I prefer to store the calculated values and get them right on the way in.

When people ask me if I prefer to put business logic in the server, it is hard to answer without a lot of information about context. But when calculations are involved the answer is yes.

The reason is that calculations are incredibly easy to fit into patterns. The patterns themselves (almost) all follow foreign keys, since the foreign key is the only way to correctly relate data between tables. So you have the "FETCH" pattern, where a price is fetched from the items table to the cart, the "EXTEND" pattern, where qty * price = extended_Price, and various "AGGREGATE" patterns, where totals are summed up to the invoice. There are others, but it is surprising how many calculations fall into these patterns.

Because these patterns are so easy to identify, it is actually conceivable to code triggers by hand to do them, but being an incurable toolmaker, I prefer to have a code generator put them together out of a data dictionary. More on that around the first of the year.

Update 1: I realize I never made it quite clear that this is part 1, as the discussion so far seems reasonable but is hardly rigorous (yet). Part 2 will be on the way after I've fattened up for the holidays.

Update 2: It is well worth following the link Mr. Koppelaars has put in the comments: http://thehelsinkideclaration.blogspot.com/2009/03/window-on-data-applications.html

Reader Dean Thrasher of Infovark has submitted a schema for review and analysis as part of my User-Submitted Analysis Request series. Today we are going to take a first look at what he has. Mr. Thrasher and I both hope that any and all readers will benefit from the exercise of publicly reviewing the schema.

This particular analysis request is a great start to the series, because it has to do with email. Everybody uses email so we all understand at a very basic level what data will be handled. Brief Introduction to User-Submitted Schemas

Mr. Thrasher and I have exchanged a couple of emails, but we have avoided any in-depth discussion. Instead, we want to carry out the conversation on the public blog. So I am not aiming to provide any "from on high" perfect analysis, instead this essay will contain a lot of questions and suggestions, and we will then move into the comments to go forward.

Disclosure: None. We are not paying each other anything, nor have I received any merchandise that would normally carry a licensing fee.

Today's essay is the very first in the User-Submitted Anlaysis Requests series. If you would like to see an analysis of your schema, follow that link and contact me.

This blog has a Complete Table of Contents and a list of Database Skills.

To get us started, I am going to quote the Infovark Product Page, and then we will see what we want to zoom in on:

Infovark automatically collects and catalogs your files and email. It consolidates your digital life into a personal wiki based on what it finds. Once you set Infovark to work, it will monitor your computer and keep your web site up-to-date

So we know even before we see anything technical that we are going to have tables of contacts, emails, phones, addresses, appointments and many other things pulled in from email systems, plus the value-add provided by the product.

We are going to start by looking at how the details of a CONTACT are stored. The schema models contacts with a group of cross references, aka many-to-many relationships, like so:

CONTACTS +----- CONTACTS-X-EMAILS -------- EMAILADDRESSES
         |
         +----- CONTACTS-X-PHONES -------- PHONES
         |
         +----- CONTACTS-X-ADDRESSES ----- ADDRESSES
         |
         +----- CONTACTS-X-WEBADDRESSES--- WEBADDRESSES

The first thing we have to note is that there is nothing wrong with this at all. It is fully normalized and so it will be very easy to make sure that database writes will not produce anomalies or bad data.

But, not surprisingly, Mr. Thrasher notes this makes for complicated SELECTS, so we want to ask if perhaps it is over-normalized, are there complications in there that do not need to be there?

If I were to follow my own advice, I would first want to identify the master tables. Master tables generally represent real things in the world: people, places, products, services, events.

So my first question is this: is an email address a free-standing entity in its own right that deserves a master table? Or is it instead a property of the CONTACT? I am going to suggest that an email address is a property of a CONTACT, and, since a CONTACT may have more than one email address, they should be stored in a child table of the CONTACTS, more like this:

CONTACTS +----- CONTACTS-X-EMAILS -------- EMAILADDRESSES
         +----- CONTACTEMAILADDRESSES
         |
         +----- CONTACTS-X-PHONES -------- PHONES
         |
         +----- CONTACTS-X-ADDRESSES ----- ADDRESSES
         |
         +----- CONTACTS-X-WEBADDRESSES--- WEBADDRESSES

Whether or not we make this switch depends not on technical arguments about keys or data types, but on whether this accurately models reality. If in fact email addresses are simply properties of contacts, then this is the simplest way to do it. Going further, the code that imports and reads the data will be easier to code, debug and maintain for two reasons: one, because it is simpler, but more importantly, two, because it accurately models reality and therefore will be easier to think about.

If this proves to be the right way to go, it may be a one-off improvement, or it may repeat itself for Phones, Addresses, and Web Addresses, but we will take that up in the next post in the series.

I am going to proceed as if this change is correct, and ask then how it will ripple through the rest of the system.

The EMAILADDRESSES table currently has these columns:

-- SQL Flavor is Firebird
CREATE TABLE EMAILADDRESS (
  ID           INTEGER NOT NULL,
  USERNAME     VARCHAR(64) NOT NULL COLLATE UNICODE_CI,
  HOSTNAME     VARCHAR(255) NOT NULL COLLATE UNICODE_CI,
  DISPLAYNAME  VARCHAR(255) NOT NULL
);

ALTER TABLE EMAILADDRESS
  ADD CONSTRAINT PK_EMAILADDRESS
  PRIMARY KEY (ID);

CREATE TRIGGER BI_EMAILADDRESS FOR EMAILADDRESS
ACTIVE BEFORE INSERT POSITION 0
AS
BEGIN
  IF (NEW.ID IS NULL) THEN
  NEW.ID = GEN_ID(GEN_EMAILADDRESS_ID,1);
END^

Suggestion: The first thing I notice is that the complete email itself is not actually stored. So we need to ask Mr. Thrasher what the thinking behind that was. My first instinct is to store that, because it is the original natural value of interest.

Suggestion: The columns USERNAME and HOSTNAME I could go either way on. If they are needed for querying and statistics, it is better to put them in. While this violates 3rd Normal Form and so puts us at risk, the values are supplied AFAIK by a batch import, and so there is only one codepath populating them, and we are likely safe. However, if we DO NOT need to query these values for statistics, and they are only there for convenience at display time, I would likely remove them and generate them on-the-fly in application code. There are some other good reasons to do this that will come up a little further along.

Suggestion: Unless I missed something in the schema sent over, we need a unique constraint on the combination of CONTACTID and USERNAME and HOSTNAME. Or, if we remove USERNAME and HOSTNAME in favor of the original EMAILADDRESS, we need a unique constraint on CONTACTID + EMAILADDRESS.

We are about to go into Part 2, which is about the other tables that reference EMAILADDRESSES, but before we do let's look at what the two tables would be if we made all changes suggested so far:

 CONTACTS             EMAILADDRESSES 
------------         --------------------
                      ID            (surrogate key)
 CONTACT_ID --------& CONTACT_ID
 other columns...     EMAILADDRESS  
                      LABEL
                      USERNAME      (possibly removed)
                      HOSTNAME      (possibly removed)
                      DISPLAYNAME

You may notice the LABEL column showed up out of nowhere. That column was previously in the cross-reference. When the cross-reference went away it landed in EMAILADDRESSES. That column LABEL holds values like "work", "home" and so on. It is supplied from whatever system we pull emails from, and so we have no constraints on it or rules about it.

Now we move on from the basic storage of EMAIL addresses to the other tables that reference those addresses. These are things like emails themselves with their lists people sent to/from, and meetings, and presumably other types of transactions as well.

When we look at transactions, which will reference contacts and email addresses, we also have to consider the fact that a CONTACT may change their email address over time. Consider a person working for Vendor A, who moves over to Vendor B. For some of the transactions they will have been at Vendor A, and then going forward they are all at Vendor B. This leads to this very important question:

Do Transactions store details about the CONTACTS as they were at the time of the transaction, or as they are now?

In other words, if a CONTACT moves from one company to another, and you look at a meeting with that person from last year, should it link to where they are now? Or should it be full of information about where they were at the time?

The answer to this question is important because it determines how to proceed on the two final points I would like to raise:

1. Should the various transactions have a foreign key back to EMAILADDRESSES, or should they simply link back to CONTACTS and contain the EMAILADDRESS itself?
2. Do we need an integer surrogate key on the EMAILADDRESSES table, especially if we do not link back to it?

So the first of the final two suggestions is: maybe the transactions tables should just link back to CONTACTID and contain a freestanding EMAILADDRESS. The first argument for this is that it preserves the history as it was, and if that is what we want, then this accomplishes it. The second argument is that by putting the actual value instead of an integer key back to some table, we simplify coding by removing a join.

The arguments against embedding the email address might be basically, "hey, if this is a kind of a data warehoues, you are really supposed to be doing the snowflake thing and you don't want to waste space on that value." To which I respond that the engineer always has the choice of trading space for speed. Putting the email in directly is a correct recording of a fact, and takes more space, but eliminates a very common JOIN from many queries, so Mr. Thrasher may choose to make that call.

This also plays back to my question about whether we should have USERNAME and HOSTNAME in the EMAILADDRESSES table. If we start putting email addresses directly into tables, we can also keep putting these other two columns in, which trades again space for speed. We could also skip them and code a parser in the application that generates them on-the-fly as needed.

Now we go all of the way back to the child table and ask a basic question: Why is there is an integer surrogate key there? Integer surrogate keys are useful in many situations, but contrary to what the web generation learned, they are not some kind of required approach in relational databases.

Consider: we need a unique constraint on CONTACTID+EMAILADDRESS anyway, so we have to justify why we would add a new column that does not add value. The reflex answer tends to be "because they join faster" but that ignores the fact that if you use the natural key of CONTACTID+EMAILADDRESS, and put these columns into child tables, you do not need to join at all! If we use the surrogate key and embed it in child tables, then getting the CONCTACT information forces two joins: through EMAILADDRESS to CONTACTS. But if we use the natural key of CONTACTID + EMAILADDRESS we already have the contact id which saves a JOIN when we are after CONTACTS details, and, unless we want to know something like LABEL, we do not have to JOIN back to EMAILADDRESSES at all. Conclusion

Well that's it. As promised, we have a few suggestions and a lot of questions for Mr. Thrasher. Check back in the coming days to see how the various questions work themselves out in the comments.

This is a follow-up to yesterday's historical perspective on ORM. In this essay we examine a particular class of business logic and ask what happens if we go server-side, embedded SQL, or ORM.

This blog has two tables of contents, the Complete Table of Contents and the list of Database Skills.

We are going to look at a process. The term is not well defined, but my own working definition is any operation that has as many of the following properties as I seem to think are important at the time I make the decision:

• Is categorically not CRUD: requires much more than displaying data to user or sending a single-row operation to the server
• Involves reading or writing many rows
• Involves reading or writing from multiple tables
• Involves multiple passes of the same data
• Involves no user interaction while executing
• If not coded to be idempotent can cause huge headaches
• Will likely take longer than a user is willing to wait (these days 3-10 seconds) and so runs in the background.
• Depends upon rules tables that control its behavior

I have a more complete description of this problem here, so this is going to be very short. The system handles magazine deliveries to stores. The shop running the system has thousands of stores and thousands of magazines. Every store has a default quantity of the particular magazines they carry. For one particular magazine, NewsTime, there are 1000 stores that get an average default quantity of 50, requiring 50,000 magazines each weak.

Here is the twist. You never get exactly 50,000, no matter what you tell the distributor. Some weeks you get 45,000, others you get 55,000, with any variation in between. So the Regulation Process adjusts the defaults for each store until the delivery amounts equal the on-hand total that was delivered on the truck.

In the first pass, we are going to consider an unrealistically simple version of Magazine Regulation, where we have too many magazines and must up the quantities until we're giving out the entire amount on-hand.

Assume a table has already been populated that has the default quantities for each store, where the relevant columns for the moment would be these:

 StoreId   |  MagazineId   |  QTY_DEFAULT | QTY_REGULATED 
-----------+---------------+--------------+---------------
    1      |      50       |      75      |      0        
    2      |      50       |      23      |      0        
    4      |      50       |      48      |      0        
   10      |      50       |      19      |      0        
   17      |      50       |     110      |      0        
   21      |      50       |      82      |      0        

We are told only that the increases must be evenly distributed, we can't just ship 5000 extra magazines to a single store. That makes sense. A simple algorithm to do this would be:

1. Give each store one additional magazine until you run out of magazines or run out of stores.
2. Repeat step 1 until you run out of magazines.

Let's mock something up in pseudo-code that shows the structure of the solution:

magsToDeliver = get total magazines...
magsDefault   = get total of defaults...

-- Outer loop implements rule 2:
-- "repeat until you run out of magazines"
while magsToDeliver > magsDefault {

   -- Inner loop implements rule 1:
   -- "increase each store by 1 until you
   --  run out of stores or run out of magazines"
   for each store getting this magazine {
        if magsToDeliver <= magsDefault break
   
        -- If you want to go nuts, and even allow
        -- the accidental simultaneous execution
        -- of two routines doing the same thing, 
        -- put these lines here instead
        magsToDeliver = get total magazines...
        magsDefault   = get total of defaults...
   
        -- This is the actual job getting done
        qty_regulate  +=1
        magsToDeliver -=1
   }
}

Let's characterize what happens with our three choices of stored procedure, embedded SQL, or ORM.

Stored Procedure. Likely the fastest solution, considering all of that row-by-row going on. If that were in app code (ORM or not) we would be making two round trips to the server per iteration.

The really crazy thing about the stored procedure solution is that it is utterly neutral to the ORM question. The entire ORM good-bad debate dissolves because there is no OOP code involved. So this could be the magic solution that ORM lovers and haters could both agree upon.

App Code With Embedded SQL (no ORM). Just about everybody hates this idea these days, but it should be here for completeness, and because there are some advantages. The top of the pseudo-code requires to aggregate pulls, and if you are not afraid of SQL you can pull down the result in one pass, instead of iterating on the client. Further, the innermost operation can be coded in SQL as a "UPDATE deliveries from (SELECT TOP 1 deliverid From deliveries...)" so that you get only one round trip per iteration, where ORM will cost two.

Any Kind of ORM. By this I mean the code will contain no SQL, and the innermost loop will likely instantiate some "delivery" objects, one after another, increment their qty_regulated property by one, and flush them out. This is twice as expensive as embedded SQL because you have to fetch the row from the database and then write it back, where the embedded SQL can issue a single command that locates and updates the row in a single statement.

Some may argue that I misunderstand ORM here, in that the library may be smart enough to allow the write without the read, and without forcing you to embed SQL. It would have to be something like A) instantiate empty object with key, B) assign value as an expression, like "+=1", C) save. I welcome any such examples and will update the post accordingly if any are forthcoming. I am assuming that no ORM tool I have seen can do this and would be happy to be corrected.

If the ORM forces us to calculate the initial sum of QTY_Default by fetching each row as an object and summing them in the app, we get an extra complete set of round trips. Bummer. But if we say, "Hey my ORM tool lets me embed SQL in *emergencies*" then perhaps we can embed a query with an aggregrate and skip that cost. But oops, we've got some embedded SQL. Leaky abstraction.

So how does it look so far? All methods have the same number of reads and writes to disk, so we are scoring them on round trips. If "X" is the number of extra magazines to be distributed, and "Y" is the number of stores getting the magazine, we have for round trips:

• Stored Procedure: 1
• Embedded SQL: X + 1 (the first pull plus one trip per extra copy of the magazine)
• ORM, Hypothetical:X + 1 (if the ORM tool can figure out how to do an update without first reading the row to the app)
• ORM, Best Case: 2X + 1 (if the first pull can be an aggregrate without embedding SQL, and two round trips per iteration)
• ORM, Worst Case:2X + Y (if the first pull must aggregate app-side and there are two round trips per iteration)

Update: if you want a laugh, check out the image on the Wikipedia page for "Business Logic", it depicts aggregation occuring on the client side.

This gives us the shape of the engineering decision. With all options reading and updating the same number of rows, it call comes down to round trips. As soon as you go client side your round trips go way up, and if your ORM tool does not support Update without Select, then it doubles from there.

Now multiply this across the entire application, every single action in code, every bit of "business logic" with any kind of loop that iterates over rows.

If you happen to know much about modern SQL, you may be aware of the amazingly cool SQL RANK() function. If this function is used in the sproc or embedded SQL approaches, you can execute the algorithm with only one loop, in a maximum of N=CEILING((Delivered-Regulated)/Stores) iterations. This will go much faster than the row-by-row, and now those two options are pulling even further ahead of the naive row-by-row methods encouraged by an ORM tool.

This ability of SQL will become extremely important, as we are about to blow apart the simplicity of the algorithm.

I have never been paid good money to write an algorithm as simple as the one described above. This is because mature businesses have always refined these simple methods for years or decades, and so the real situation is always more complex.

In a real magazine regulation algorithm, the rules tend to be more like this:

1. Apply all of these rules whether you are increasing or decreasing the amounts to deliver
2. Stop applying the rules when delivery amounts have been balanced to what we have on hand, no matter where you are in the process
3. Always increase/decrease any particular store by exactly one on each iteration, no matter which rule you are working on
4. Never decrease any store below 2
5. Decrease any store whose past 3 issues sold less than 60% by 1, unless this would project their sales of this issue above 60%, and prioritize by prior sales % ascending.
6. If the previous step completes, and we are short of magazines decrease each store by 1 by order of previous sales percents ascending. Repeat until we are in balance.
7. If all stores are reduced to 2 and we are still short, abort and write error to log.
8. If after the decreases we have excess magazines, increase any store whose past 3 issues sold more than 70% by 1, unless this would reduce their projected sales of this issue below 70%, and prioritize by prior sales % descending (so the stores with the most sales are handled first in case we don't get to all of them)
9. If the previous step completes, and we are still in excess, increase each store by 1 in order of previous sales percents descending. Repeat until we are in balance.

This can also get doubled again if we must implement one set of rules when we start out with a too few magazines, and another set of rules when we start out with too many.

Well, it's not that hard. It actually comes down to having four outer loops in succession. The percent-based reduction, then the by 1 reduction, then the percent-based increase, then the by 1 increase.

But technically the more important matters are these:

• We now have to grab the sales % for every store for this magazine on their past 3 issues and keep it handy throughout the routine.
• The rules stated above contain constants like 70%, 60%. These would properly be in some parameter table to allow the user to control them, so those have to be fetched.
• The loop through the stores is now much different, as we are filtering on prior sales percent for the percent-based passes, and filtering and ordering on prior sales percent for the by 1 passes.

Now let's see how our three approaches would change.

The Improved Stored Procedure. If we change the sproc to use RANK() and make batch updates, we would pull the prior sales percents into a temp table and apply a covering index to cut the reads from that table in half. Our main loop would then simply join to this temp table and use it for both filtering and ordering.

The Embedded SQL. If we also changed the embedded SQL so it was making batch updates with RANK(), we would also generate a temp table. This option remains the same as the sproc except for where we put the SQL. However, it now has far fewer round trips, and starts to look much more like the sproc in terms of performance.

The ORM Approach. The idea here would be to get those prior sales percents down into an ordered collection and then use them as the basis for the loops. The thorny part is that they must be aggregated and sorted. If we want to avoid all embedded SQL, then the aggregation can be done client-side if we don't mind pulling down 3 times as many rows as are required. The sorting we can pull off if we put the objects into a collection such as an associative array, where the key is the sales percent, then we can use [language of choice]'s built-in sorting (hopefully), and we have escaped the dread evil of embedded SQL.

So we end up where we were, only more so. The sproc remains the fastest, and if we know how to code set-oriented nifty stuff with RANK() then the embedded SQL will run in almost the exact same time. The ORM requires most likely even more round trips and expensive app-side operations that are performed much more efficiently in the db server, unless we are willing to break the abstraction and embed a bit of SQL.

But in the end, if all of that cost of the ORM kicks a 3 second routine to 7 seconds, still well below what any user would notice, and you avoid embedded SQL, and it lets you keep your paradigm, who am I to judge?

I offer none. There are so many conventions in play regarding where to put your code, what tool you are already using, and so forth, that it is really up to the reader to draw conclusions. I only hope there is enough information here to do so.

A couple of years ago I broke my basic rule of sticking to practical how-to and general programming philosophy and wrote Why I Do Not Use ORM. It sure got a lot of hits, and is read every day by people searching such things as "orm bad" or "why use orm". But I have never been satisfied with that post, and so I decided to take another stab from another angle. There are legitimate problems that led to ORM, and those problems need to be looked at even if we cannot quite agree on what they are or if ORM is the answer.

UPDATE: In response to comments below and on reddit.com, I have a new post that gives a detailed analysis of an algorithm implemented as a sproc, in app code with embedded SQL, and in ORM.

Here then, is one man's short history of commercial database application programming, from long before the ORM system, right up to the present.

This blog has two tables of contents, the Topical Table of Contents and the list of Database Skills.

When I began my career the world was a different place. No Web, no Java, and Object Orientation had not yet entered the mainstream. My first application was written on a timeshare system (a microVAX) and writing LAN applications made me a good living for awhile before I graduated to client/server.

In those days there were three things a programmer (We were not "software engineers" yet, just programmers) had to know. Every programmer I knew wanted to master all of these skills. They were:

• How to design a database schema for correctness and efficiency.
• How to code an application that could process data from the database, correctly and efficiently.
• How to make a good UI, which came down to hotkeys and stuffing the screen with as much info as possible.

In this essay we are going to look at those first two.

My own experience may be somewhat peculiar in that I have never worked on a team where the programmers were separated from the database. (OK, one exception, in my current assignment there is an iron curtain between the two, but happily it is not my problem from where I sit). Coders made tables, and "tablers" wrote code. So this focus on being a good developer by developing both skills may be rare, enjoyed by those who have the same ecumenical background that I enjoyed.

Things changed rapidly, but most of those changes did not really affect application development.

When Windows 95 came out, being "almost as good as a Mac", we recoded our DOS apps into Windows apps without too much trouble and life went on as before.

Laser printers replaced dot-matrix for most office use, CPUs kept getting faster (and Windows kept getting slower), each year there were more colors on the screen, disks got bigger and RAM got cheaper.

Only the internet and the new stateless programming required any real adjustment, but it was easy for a database guy because good practice had always been to keep your transactions as short as possible. The stateless thing just kind of tuned that to a sharp edge.

Finally, with the internet, the RDBMS finally lost its place as sole king of the datastore realm, but those new datastores will have to wait for another day, lest we get bogged down.

Arguably nothing changed programming more than Object Orientation. Certainly not Windows 95, faster graphics or any of those other Moore's Law consequences. I would go so far as to say that even the explosion of the web just produced more programming, and of different kinds of apps, and even that did not come close to the impact of Object Orientation. Disagree if you like, but as it came in, it was new, it was strange, it was beautiful, and we were in love.

Now here is something you may not believe. The biggest question for those of us already successfully developing large applications was: What is it good for? What does it give me that I do not already have? Sure its beautiful, but what does it do?

User interfaces were for me the easiest first place to see the benefits. When the widgets became classes and objects, and we empolyed encapsulation, inheritance and composition, the world changed and I don't know anybody who ever looked back.

But in the matter of processing data, things were not so clear cut. The biggest reason may have been that all languages back then had specialized data structures that were highly tuned to handling relational data. These worked so well that nobody at first envisioned anything like ActiveRecord because we just did not need it.

With these structures you could write applications that ran processes involving dozens of tables, lasting hours, and never wonder, "Gosh, how do I map this data to my language of choice?" You chose the language you were using precisely because it knew how to handle data!

I would like to throw in just one example to show how OOP was not relevant to getting work done back then. I was once asked to optimize something called "ERP Allocation" that ran once/day, but was taking 26 hours at the largest customer site, obviously a big problem. It turned out there was a call to the database inside of a tightly nested loop, and when I moved the query outside of the loop the results were dramatic. The programmers got the idea and they took over from there. The main point of course is that it was all about how to efficiently use a database. The language was OOP, and the code was in a class, but that had nothing to do with the problem or the solution. Going further, coding a process so data intensive as this one using ActiveRecord was prima facia absurd to anybody who knew about data and code.

But the web had another impact that was far more important than just switching to stateless programming. This was the introduction of an entirely new family of languages that took over the application space, listed here in no particular order: Perl, PHP, Python, Ruby, and the king of them all: Java.

All of these languages have one thing in common that positively jumps out at a veteran: a complete lack of data structures specialized for handling relational data. So as these languages exploded in popularity with their dismal offerings in data handling, the need to provide something better in that area became rapidly clear.

Java has a special role to play because it was pure OOP from the ground up. Even the whitespace is an object! The impact of Java is very important here because Object Orientation was now the One True Faith, and languages with a more flexible approach were gradually demoted to mere 'scripting' languages. ( Of course proponents will quickly point out that 1/12 of the world's population is now using a single application written in one of those 'scripting' languages).

So the explosion of languages without decent data handling abilities, coupled with a rise in OOP-uber-alles thinking led us quite naturally to:

The first premise of ORM is that there is a design mismatch between OOP and Relational, which must resolved before any meaningful work can be done.

This view is easy to sympathize with, even if you disagree, when you consider the points raised in the above sections, that the languages in play lack any real specialized data structures, and that a certain exclusive truthiness to OOP has arisen that is blind to entire classes of solutions.

So we must grant the ORM crowd their first premise, in modified form. It is not that there is a design mismatch, it is that there is something missing, something that was in older systems that is just not there in the newer languages. Granting that this missing feature is an actual mismatch requires a belief in the Exclusive Truth of OOP, which I do not grant. OOP is like the computer itself, of which Commander Spock said, "Computers make excellent servants, but I have no wish to be servant to a computer."

But anyway, getting back to the story, the race was on to replace what had been lost, and to do it in an OOPy way.

Fast forward and we soon have an entire family of tools known as Object-Relational-Mappers, or ORM. With them came an old idea: persistence.

The idea has always been around that databases exist to persist the work of the programmer. I thought that myself when I was, oh, about 25 or so. I learned fast that my view of reality was, *cough*, lacking, and that in fact there are two things that are truly real for a developer:

• The users, who create the paycheck, and
• The data, which those users seemed to think was supposed to be correct 100% of the time.

From this perspective, the application code suddenly becomes a go-between, the necessary appliance that gets data from the db to the user (who creates the paycheck), and takes instructions back from the user and puts them in the database (correctly, thank you, and don't make the user wait). No matter how beautiful the code was, the user would only ever see the screen (or page nowadays) and you only heard about it if it was wrong. Nobody cares about my code, nobody cares about yours.

However, in the ORM world the idea of a database as the persistence layer now sits on a throne reserved for axiomatic truth. Those who disagree with me on this may say that I have the mistaken perspective of an outsider, to which I could say only that it is this very idea that keeps me an outsider.

But we should not paint the world with a broad brush. Chris Wong writes an excellent blog where he occassionally details how to respect the database while using Hibernate, in this post and this post.

There are plenty of alternatives to ORM, but I would contend that they begin with a different world view. Good business recognizes the infinite value of the users as the generators of the Almighty Paycheck, and the database as the permanent record of a job well done.

This worldview forces us into a humble position with respect to our own application code, which is that it is little more than a waiter, carrying orders to the kitchen and food back to the patrons. When we see it this way, the goal becomes to write code that can efficiently get data back and forth. A small handful of library routines can trap SQL injection, validate types, and ship data off to the database. Another set can generate HTML, or, can simply pass JSON data up to those nifty browser client libraries like ExtJS (now "Sencha" for some reason).

This covers a huge amount of what an application does, if you do not have much in the way of business logic.

I have an entire essay on this about half-written, but in short, it comes down to understanding what business logic really is. Update: This post is now available

The tables themselves are the bottom layer of business logic. The table design itself implements the foundation for all of the business rules. This is why it is so important to get it right. The tables are organized using normalization to have a place for everything and everything in its place, and after that the application code mostly writes itself.

The application code then falls into two areas: value-add and no value-add. There is no value-add when the application simply ships data off to the user or executes a user request to update the database. Those kinds of things should be handled with the lightest possible library that gets the job done.

But the value-add stuff is different, where a user's request requires lookups, possibly computations and so forth. The problem here is that a naive analysis of requirements (particulary the transliteration error (Scroll down to "The Customer Does Not Design Tables) will tend to generate many cases of perceived need for value-add where a simpler design can reduce these cases to no value-add. But even when the database has been simplified to pristine perfection, there are jobs that require loops, multiple passes and so forth, which must be made idempotent and robust, which will always require some extra coding. But if you know what you are doing, these always turn out to be the ERP Allocation example given above: they are a lot more about the data than the classes.

Another huge factor is where you come down on the normalization debate, particularly on the inclusion of derived values. If you keep derived values out of the database, which is technically correct from a limited perspective, then suddenly the value-add code is much more important because without it your data is incomplete. If you elect to put derived values into your database than value-add code is only required when writing to the database, so huge abstractions meant to handle any read/write situation are unnecessary. (And of course, it is extremely important to Keep denormalized values correct ).

This essay hardly covers the entirety of making code and data work together. You still have to synchronize schema changes to code, and I still think a data dictionary is the best D-R-Y way to do that.

I hope this essay shows something of why many programmers are so down on ORM, but much more importantly that there are coherent philosophies out there that begin with a different worldview and deliver what we were all doing before ORM and what we will all still be doing after ORM: delivering data back and forth between user and database.

A database is not much without the applications that connect to it, and one of the most important factors that affects the application's performance is how it retrieves data from queries. In this essay we are going to see the effect of round trips on application performance.

This blog has two tables of contents, the Complete Table of Contents and the list of Database Skills.

The test will pull 15,000 rows from a table. We do it three different ways and see which is faster and by how much.

The script below creates a table and puts 1 million rows into it. We want far more rows in the table than we will actually pull so that we can pull fresh rows on every pass through the test. It is deliberately crafted to spread out the adjacent values of the integer primary key. This is because, inasmuch as can control what is going on, we want every single row to be on a different page, so that in all tests the cost of retrieving the row is roughly the same and we are measuring only the effect of our retrieval methods.

The script can be run without modification in pgAdmin3, and with slight mods on MS SQL Server.

create table test000 (
    intpk int primary key
   ,filler char(40)
)


--  BLOCK 1, first 5000 rows
--  pgAdmin3: run as pgScript
--  All others: modify as required  
--
declare @x,@y;
set @x = 1;
set @y = string(40,40,1);
while @x <= 5000 begin
    insert into test000 (intpk,filler)
    values ((@x-1)*200 +1,'@y');

    set @x = @x + 1;
end

-- BLOCK 2, put 5000 rows aside 
--
select  * into test000_temp from test000

-- BLOCK 3, Insert the 5000 rows 199 more
--          times to get 1million altogether
--  pgAdmin3: run as pgScript
--  All others: modify as required  
--  
declare @x;
set @x = 1;
while @x <= 199 begin
    insert into test000 (intpk,filler)
    select intpk+@x,filler from test000_temp;

    set @x = @x + 1;
end

The simplest code is a straight loop that pulls 15,000 consecutive rows by sending an explicit query for each one.

# Make a database connection
$dbConn = pg_connect("dbname=roundTrips user=postgres");

# Program 1, Individual explicit fetches
$x1 = rand(0,199)*5000 + 1;
$x2 = $x1 + 14999;
echo "\nTest 1, using $x1 to $x2";
$timeBegin = microtime(true);
while ($x1++ <= $x2) {
    $dbResult = pg_exec("select * from test000 where intpk=$x1");
    $row = pg_fetch_array($dbResult);
}
$elapsed = microtime(true)-$timeBegin;
echo "\nTest 1, elapsed time: ".$elapsed;
echo "\n";

The next command asks the server to prepare a statement, but it still makes 15,000 round trips, executing the prepared statement with a new parameter each time. The code looks like this:

# Make a database connection
$dbConn = pg_connect("dbname=roundTrips user=postgres");

# Program 2, Individual fetches with prepared statements
$x1 = rand(0,199)*5000 + 1;
$x2 = $x1 + 14999;
echo "\nTest 2, using $x1 to $x2";
$timeBegin = microtime(true);
$dbResult = pg_prepare("test000","select * from test000 where intpk=$1");
while ($x1++ <= $x2) {
    $pqResult = pg_execute("test000",array($x1));
    $row = pg_fetch_all($pqResult);
}
$elapsed = microtime(true)-$timeBegin;
echo "\nTest 2, elapsed time: ".$elapsed;
echo "\n";

This time we issue a single command to retrieve 15,000 rows, then we pull them all down in one shot.

# Make a database connection
$dbConn = pg_connect("dbname=roundTrips user=postgres");

# Program 3, One fetch, pull all rows
$timeBegin = microtime(true);
$x1 = rand(0,199)*5000 + 1;
$x2 = $x1 + 14999;
echo "\nTest 3, using $x1 to $x2";
$dbResult = pg_exec(
    "select * from test000 where intpk between $x1 and $x2"
);
$allRows = pg_fetch_all($dbResult);
$elapsed = microtime(true)-$timeBegin;
echo "\nTest 3, elapsed time: ".$elapsed;
echo "\n";

I ran this five times in a row, and this is what I got:

Compared to the naive example, the set-oriented fetch of al 15,000 rows in a single shot ran 40 times faster. This is what set-oriented code does for an application.

While the prepared statement option ran faster than the naive option, the set oriented example still ran 25 times faster than the repeated prepared statements.

I also re-arranged the order of the tests, and the results were the same.

So this test was done using PHP against PostgreSQL, will other servers and client languages get different results? Given the same hardware, a different client language or server is going to have a different spread but the shape will be the same. Fetching all rows in a single shot beats the living frack out of round trips inside of loops in any client language against any server.

The most obvious conclusion is that any query returning more than 1 row should return all rows as a set. The advantage is so stark with large row counts that it is worthwhile making this the default for our applications, unless we can find a very good reason not to. So what would the objections be?

One objection might go something like, "Ken, I see the numbers, but I know my app very well and we never pull more than 10-20 rows in a pop. I cannot imagine how it would matter at 10-20 rows, and I do not want to recode." This makes sense so I ran a few more tests with 20 and 100 rows, and found that, on my hardware, you need about 100 rows to see a difference. At 20 rows all three are neck-in-neck and at 100 the set is pulling 4 times faster than the prepared statement and 6 times faster than the naive statement. So the conclusion is not an absolute after all, some judgment is in order.

Another thing to consider is how many simultaneous reads and writes might be going on at any given time. If your system is known to have simultaneous transactions running regularly, then the complete fetch may be a good idea even if you do some tests for best-guess row count and the tests are inconclusive. The reason is that the test is a single user case, but multiple simultaneous users put a strain on the database, even when they are not accessing the same tables. In this case we want the application to play the "good citizen" and get in and out as quickly as possible to reduce strain on the server, which will improve the performance of the entire application, not just the portions optimized for complete fetches.

Another objection might be, "Well, my code needs to pull from multiple tables, so I cannot really do this. When we do -PROCESS-X- we go row by row and need to pull from multiple tables for each row." In this case you *definitely* need to go set oriented and pull all associated quantities down in a query with a JOIN or two. Consider this, if on your particular hardware the ratio of naive row-by-row to single fetch is 10, and you must pull from 2 other tables for each row, that means you are really running 30 times slower (ratio is 10 x 3 reads) than you could be.

Back when dinosaurs ruled the Earth and there was no internet (outside of Universities, etc), the languages we used had specialized data structures that were tuned to database use. Compared to those older systems the newer languages born on the internet are more or less starving for such a data structure.

PHP gets by fairly well because its associative array can be used as a passive (non object-oriented) data structure that comes pretty close to what we had before.

I bring this up because the choice of a language and its support for a "fetch all" operation obviously impacts how well the conclusions of this post can be implemented. If your mapping tool has an iterator that absolves you of all knowledge of what is going on under the hood, it may be worthwhile to see if it is doing a complete fetch or a row-by-row.

I generally do not reveal too many details about systems I design for customers or employers. This leaves me sometimes in a bind for example material. I either have to simplify it beyond what I would like, or make something up that I have not actually put into Production.

On top of that, one of the key themes of this blog is that table design is a crucial skill, and if the examples I give do not match what you are doing, they may be hard to make use of.

So I would like invite analysis requests. Go over to the Contact the Author page and drop me an email and tell me about the system you are trying to design or optimize.

There are no rules on the type of system.

The most interesting mini-projects would be those where advice you have been given elsewhere (or here for that matter) does not seem to fit.

I will do my best to reply, even if I have to say no, so that nobody is left wondering.

Remember this blog is one of those hobby/professional things, good for all of us but nobody is getting paid, so if you are in a terrible hurry this might not be the best thing.

Good table design is a great foundation for a successful application stack. Table design patterns basically resolve into master tables and transaction tables. When we know a thing or two about the master tables (or entities if you prefer), we can infer a great deal about the transactions.

This blog has two tables of contents, the Topical Table of Contents and the list of Database Skills.

Imagine we have been asked to recode the company's time-billing system. Because this is for the company we work for, we have some inside knowledge about how things work. We know that:

• There are, of course, customers.
• ....and employees who record time
• Each hour we record goes against a Work Order
• There are different kinds of jobs, like project management, programming, programming management, and others.

Knowing only this, is it possible to anticipate what the system will look like? A safe answer is "no", on the claim that we will undoubtedly learn more, but this safe answer happens to be wrong. We can in fact anticipate the overall shape of the system, and new information will shift details, but it will not change the shape.

We can anticipate the nature of the transactions if we determine the upper bound of complexity and the combinatorial completeness of the system.

We can safely assume that the big number to get right is going to be the billing rate. Our employer assumes we will get everything else right, but the billing rate is going to have them chewing their fingernails until they know we understand it and have coded it correctly.

The cool thing is that we already have enough information to establish an upper bound on the complexity of the system by looking at the master tables, where a master table is generally one that lists details about real things like people, places, things, or activities. So far we know (or think we know) about three master tables:

• Customers
• Employees
• Services

Now we define the upper bound of complexity as:

The upper bound of complexity occurs when the billing rate is determined by all three master entities.

In plain English, calculating a billing rate can be as complicated as looking up a rate specific to a customer for a service for an employee but cannot be more complex than that because there are no other entities with which to work.

We can also anticipate all possible calculations for the billing rate by working through the complete set of combinations of master entities. This would look like the list below. Note that we are not trying to figure out right now which of these is likely to occur, we just want to get them listed out:

• Each service has a default rate
• Each customer has a negotiated rate
• Each employee bills out at a default rate
• The combination customer-service may have a rate
• The combination customer-employee may have a rate
• The combination customer-service-employee may have a rate (this is the upper bound of complexity, all three master entities determine the rate).

Unless we live in a super-simple world where only the first item in the list is present, we will end up dealing with several if not all of the combinations listed above.

Each of these combinations then becomes a table, and we know the billing rate will be determined by a resolution. New Information

Now comes the big day and we interview with somebody we'll call "The Explainer" who is going to officially explain the billing system. Can he break what we already know? No. At most he can:

• Make us aware of new master entities, perhaps there are "projects" and "contracts" that get their own billing arrangements.
• Dispel our notions about some of the combinations by saying, "Oh we never give a customer a default rate, the default rates come out of the services."

What about the case where we know absolutely nothing about an assignment when we go in to begin the interviews? We can do a good job of thinking on our feet if we draw "The Explainer" towards the master entities. As we gain confidence that we know what the master entities are, we can ask questions to probe Combinatorial Completeness and the Upper Bound of Complexity.

One caveat: This method works for transactions between master entities. When "The Explainer" starts describing something that cannot be recognized as an interaction between master entities, do not try to stuff the problem into this box, it may not fit.

At this point, we can also anticipate a lot of what the application will look like. We will need maintenance screens for all of the master entities, and a really slick UI will allow for very easy editing of those various cross-reference combination tables. As long as that much is done, we are almost finished, but not yet.

There will be some billing process that pulls the time entries, finds the correct billing rate for each one, and permanently records the invoices. If we use a resolution this task is child's play to code, debug, and maintain.

Then of course there is the presentation, the actual bill. Depending on the company, these may be delivered as hardcopy or in email. That will of course have to be coded up.

There are two conclusions. First, as originally stated, many transactions can be anticipated when you know what the master entities are.

But secondly, and every bit as important, once the table design is sound, the application pretty much writes itself. On a personal note, this is probably why I do not find application coding as exciting as I once did. Once I realized that the real challenge and satisfaction was in working out the tables, the coding of the app became a bit of a drudge, it requires no judgment as far as business rules are concerned.

If you regularly code queries and have never been introduced to the windowing functions, then you are in for a treat. I've been meaning to write about these for over a year, and now it's time to get down to it.

SQL Server calls these functions Ranking Functions.

PostgreSQL supports a wider range of functions than MS SQL Server, having put them in at 8.4, and PostgreSQL and calls them Window Functions.

Oracle's support is broader (by a reading of the docs) than SQL Server or PostgreSQL, and they call them Analytic Functions.

I try to stay away from MySQL, but I did a quick Google on all three terms and came up with a few forum posts asking when and if they will be supported.

In this post we are going to look at NTILE, a cool function that allows you to segment query results into groups and put numbers onto them. The name is easy to remember because it can create any -tile, a percentile, a decile, or anything else. In short, an n-tile. But it is much easier to understand with an example, so let's go right to it.

Consider a table of completed sales, perhaps on an eCommerce site. The Sales Manager would like them divided up into quartiles, four equally divided groups, and she wants the average and maximum sale in each quartile. Let's say the company is not exactly hopping, and there are only twelve sales, which is good because we can list them all for the example. If we already had the quartiles provided then the query would be easy, so if we were lucky enough to be starting with this:

 CUSTTYPE | AMOUNT  | QUARTILE
----------+---------+----------
 RETAIL   |   78.00 |   1
 RETAIL   |  234.00 |   1
 DEALER   |  249.00 |   1
 DEALER   |  278.00 |   2
 RETAIL   |  392.00 |   2
 RETAIL   |  498.00 |   2
 DEALER   |  500.00 |   3
 RETAIL   |  738.00 |   3
 DEALER   | 1250.00 |   3
 RETAIL   | 2029.00 |   4
 RETAIL   | 2393.00 |   4
 RETAIL   | 3933.00 |   4

The query would be child's play if we already had the quartile:

Select quartile
     , avg(amount) as avgAmount
     , max(amount) as maxAmount
  FROM ORDERS
 GROUP BY quartile
 ORDER BY quartile

The problem of course is that we do not usually have handy columns like QUARTILE provided, but we can generate the QUARTILE column during the query by using NTILE.

Select quartile
     , avg(amount) as avgAmount
     , max(amount) as maxAmount
  FROM (
        -- The subquery is necessary
        -- to process all rows and add the quartile column
        SELECT amount
             , ntile(4) over (order by amount) as quartile
          FROM ORDERS
       ) x
 GROUP BY quartile
 ORDER BY quartile

This query will give us what the Sales Manager wants.

The NTILE() function takes a single argument, which tells the server how many groups to divide the data into. If there are not an exact number of rows in each group, the server decides which groups will be missing one row. So in an exact case all of your groups have the same count of rows, but when it does not divide evenly, one or more of them will be one row short.

If you pass 100 to NTILE(), you get a percentile. If you pass 10, you get a decile, and so forth.

The magic is in the OVER() function. This supports two clauses, and the example shows one, the ORDER BY. Quite simply, the ORDER BY clause tells the server how to line up the rows when adding the NTILE values. The clause is very flexible, and has nothing to do with your query's overall ORDER BY clause. The Second Clause: PARTITION

Now we will pretend the Sales Manager is not satisfied, and wants separate numbers for the two Customer Types. We could do this if the NTILE() function would create two sets of quartiles, one for each Customer Type, like so:

 CUSTTYPE | AMOUNT  | QUARTILE
----------+---------+----------
 DEALER   |  249.00 |   1
 DEALER   |  278.00 |   2
 DEALER   |  500.00 |   3
 DEALER   | 1250.00 |   4
 RETAIL   |   78.00 |   1
 RETAIL   |  234.00 |   1
 RETAIL   |  392.00 |   2
 RETAIL   |  498.00 |   2
 RETAIL   |  738.00 |   3
 RETAIL   | 2029.00 |   3
 RETAIL   | 2393.00 |   4
 RETAIL   | 3933.00 |   4

We can do this by using the PARTITION BY clause, which tells the server to break the rows into groups and apply the NTILE() numbering separately within each group. The new query would be this:

Select custtype
     , quartile
     , avg(amount) as avgAmount
     , max(amount) as maxAmount
  FROM (
        -- The subquery is necessary
        -- to process all rows and add the quartile column
        SELECT amount
             , ntile(4) over (partition by custtype
                                 order by amount) as quartile
          FROM ORDERS
       ) x
 GROUP BY custtype,quartile
 ORDER BY custtype,quartile

Now once again the Sales Manager, who is never satisified, comes down and says that the average is no good, she needs the max and the median sale value within each quartile. To keep it simple, she does not need this broken out by customer type, it can be applied to the entire set.

This is a case where we can use NTILE() twice. The first time we will break all sales up into four groups, to get the quartiles, and then we will break up each quartile into two groups to get the median. The code looks like this:

Select quartile
     , max(case when bitile=1 then amount else 0 end) as medAmount
     , max(amount) as maxAmount
  FROM (
        -- The second pass adds the
        -- 2-tile value we will use to find medians
        SELECT quartile
             , amount
             , ntile(2) over (partition by quartile
                                  order by amount) as bitile
          FROM (
                -- The subquery is necessary
                -- to process all rows and add the quartile column
                SELECT amount
                     , ntile(4) over (order by amount) as quartile
                  FROM ORDERS
               ) x1
       ) x2
 GROUP BY quartile
 ORDER BY quartile

The magic here is that we know we've divided the data evenly into four sets, so the median will be the maximum value half way through each set. In other words, it will be the maximum value when the value of bitile=1 for each quartile.

Once you get down the basics of the OVER clause, Oracle looks really good, because they support the clause over the largest range of functions, at least going by the respective doc pages for each platform.

Sometimes you need to process a table row-by-row, and the established approach is to use cursors, which are verbose, slow, and painful to code and use.

Here is the basic minimum syntax required to loop through a table and get something done. The SQL flavor is MS SQL Server, but its not much better in any other flavor.

-- I coded this off the top of my head, there
-- may be a minor syntax error or two

-- Most of this is pseudo-code, but take
-- note that it is ordered on column1
declare someCursorName cursor for
 select column1, column2, column3 
   from anyTable
  ORDER BY column1

-- Have to do this now
open someCursorName

-- Now you need to declare some variables
-- For the example I'm just making everything int
declare @column1 int
      , @column2 int
      , @column3 int

-- Gosh, we're actually about to start the loop!  Finally!
fetch next from someCursorName into @column1,@column2,@column3
while @@fetch_status = 0 begin

   --  If you still remember what you actually wanted
   --  to do inside the loop, code it here:

-- Repeat this line from the top here again:
fetch next from someCursorName into @column1,@column2,@column3
end

-- Not done yet, these two lines are crucial
close someCursorName
deallocate someCursorName

Call me petty, but what I hate about that code is that I have to refer to specific columns of interest 3 times (not counting the declarations). You refer to them in the cursor declaration and in the two FETCH commands. With a little clever coding, we can vastly simplify this and do it only once.

We can execute the same loop without the cursor if one of the columns is ordered and unique. Let us say that column1 is the primary key, and is an auto-incremented integer. So it is ordered and unique. The code now collapses down to:

-- I coded this off the top of my head, there
-- may be a minor syntax error or two

-- We can't get around declaring the vars, so do that
declare @column1 int
      , @column2 int
      , @column3 int

-- If you know a safe value for initialization, you
-- can use the code below.  If this is not 100% 
-- safe, you must query for the value or it must
-- be supplied from some other source
set @column1 = -1

-- BONUS POINTS: Can this become an infinite loop?
while 1 = 1 begin

-- Now we code the query and exit condition
 select TOP 1
        @column1 = column1
      , @column2 = column2
      , @column3 = column3 
   from anyTable
  WHERE column1 > @column1  -- this is what advances the loop
  ORDER BY column1

if @@rowcount = 0 begin
    break
end

        -- Put the actions here        

end

The only requirement for this approach is that you have a unique ordered column. This usually means a unique key or primary key. If "column1" is not unique, the loop will skip all but the first value in each group.

Also, it is very nice if you know a safe value to use as an initializer. Without that, you must query for the minimum value that matches the condition and then decrement it by one.

Finally, can this loop become infinite? No. Well, if, in the extremely unlikely situation that rows are being added to the base table faster than you are processing them, then yes, it could go on for a very long time. But if that were happening I'd say there was a separate problem to look at.

It should probably go without saying, but if the particular loop is going to happen very often, the table should be indexed on your unique ordered column. If it is a primary key or you already have a unique constraint it is not necessary to create an index explicitly because there will be one as part of the key or constraint.

In this blog I have done at many articles on Normalization and Denormalization, but I have never put all of the arguments together in one place, so that is what I would like to do today.

There are links to related essays on normalization and denormalization at the bottom of this post.

This blog has two tables of contents, the Topical Table of Contents and the list of Database Skills.

Normalization is the process of designing tables so that each fact is stored in exactly one place. A "fact" in this case is any detail that we have to keep track of, such as a product's description, a product's price, an employee's social security number, and so forth.

The process is all about figuring out what tables you need and what columns each table will have. If we are talking about an employee's social security number, then we can guess right from the start that will have a table of EMPLOYEES, and that one of the columns will be SSN. As we get more details, we add more tables and columns.

The advantage of normalization comes when your application writes data to the database. In the simplest terms, when the application needs to store some fact, it only has to go to one place to do it. Writing this kind of code is very easy. Easy to write, easy to debug, easy to maintain and improve.

When the database is not normalized, you end up spending more time writing more complicated application code that is harder to debug. The chances of bad data in your production database go way up. When a shop first experiences bad data in production, it starts to become tempting to "lock down" access to the database, either by forcing updates to go through stored procedures or by trying to enforce access to certain tables through certain codepaths. Both of these strategies: stored procedures and code paths, are the actually the same strategy implemented in different tiers, they both try to prevent bugs by routing access through some bit of code that "knows what to do." But if the database is normalized, you do not need any magic code that "knows what to do."

So that, in brief, is what normalization is and why we do it. Let's move on now to denormalization.

Normalization is easy to explain because there is a clearly stated end-goal: correct data. Moreover, there are well-defined methods for reaching the goal, which we call the normal forms, First Normal Form, Second Normal Form, and higher forms. By contrast, denormalization is much harder to talk about because there is no agreed-upon end goal. To make matters worse, denormalization violates the original theory of Relational Databases, so you still have plenty of people screaming not to do it all, making things even more confusing. What we have now in our industry is different shops denormalizing in different ways for different reasons.

The arguments that I have heard in my career boil down to two basic groups. The first set of arguments centers around calculated or derived values, and the second set centers around programmer convenience.

My own experience comes down heavily in favor of denormalizing by storing derived values directly into the tables, with the extremely signficant caveat that you must have a way to ensure that they are always correct. In this paradigm you maintain strict normalization for facts supplied from the outside, and then layer on additional facts that are calculated during write operations and saved permanently.

Here is a very simple example. A strictly normalized database happens to be missing data that many programmers would automatically assume should be stored. Believe it or not, a simple value in a shopping cart like EXTENDED_PRICE is forbidden by 3rd normal form because it is a non-key dependency, or, in plain English, since it can be derived from other values (QTY * PRICE), then it is redundant, and we no longer have each fact stored in exactly one place. The value of EXTENDED_PRICE is only correct if it always equals QTY * PRICE, and so there is now a "fact" that is spread across three locations. If you store EXTENDED_PRICE, but do not have a way to ensure that it will always 100% of the time equal QTY * PRICE, then you will get bad data.

So, given the risk of bad data, what is to be gained by putting EXTENDED_PRICE into the cart? The answer is that it adds value to the database and actually simplifies application code. To see why, imagine a simple eCommerce shopping cart that does not store any derived values. Every single display of the cart to the user must go all over the place to gather lots of details and recalculate everything. This means re-calculating not just the EXTENDED_PRICE, but adding in item level discounts, taking account of possible tax exemptions for different items, rolling the totals to the cart, adding in tax, shipping, perhaps a customer discount, a coupon, and who knows what else. All of this just to display the cart, every time, no matter what the purpose.

This situation leads to three problems. A pitifully slow application (too many disk reads and lots of cycles calculating the values), maddening bugs when an application update has subtle changes to the calculations so the customer's order no longer displays the same numbers as it did yesterday, and the frustrating requirement that the simplest of reports must route through application code to calculate these values instead of simply reading them off the disk, which leads to reporting systems that are orders of magnitude slower than they could be and horribly more complicated than they need to be because they can't just read straight from the tables.

Now let's look at how that same shopping cart would be used if all of those calculated values were generated and saved when the order is written. Building on your foundation of normalized values (price, qty), you need only one body of code that has to perform calculations. This magic body of code takes the user-supplied values, adds in the calculations, and commits the changes. All other subsequent operations need only to read and display the data, making them faster, simpler, and more robust.

So the obvious question is how to make sure the derived values are correct. If they are correct, we gain the benefits with no down side. If there is the smallest chance of bad data, we will quickly pay back any benefit we gained by chasing down the mistakes.

From a technical standpoint, what we really need is some technology that will make sure the calculations cannot be subverted, it cannot be possible for a stray bit of program code or SQL Statement to put the wrong value in for EXTENDED_PRICE. There are a few generally accepted ways to do this:

• Require all writes to go through a certain codepath. The only PRO here is that you keep the logic in the application code, and since most shops have more programmers than database people, this makes sense. The only CON is that it never works. One programmer working alone can maintain discipline, but a team cannot. All it takes is one programmer who did not know about the required codepath to screw it all up. Also, it makes your system inflexible, as it is no longer safe to write to the database except through a single application.
• Require all writes to go through stored procedures. This is nominally better than the codepath solution because it is not subvertible, and you can allow different side apps and utilities to safely write to the database. But it makes a lot of work and tends to be very inflexible.
• Putting triggers onto tables that perform the calculations and throw errors if a SQL statement attempts to explicitly write to a derived column. This makes the values completely non-subvertible, ensures they will always be correct, and allows access from any application or utility. The downside is that the triggers cannot be coded by hand except at extreme cost, and so must be generated from a data dictionary, which is fairly easy to do but tends to involve extreme psychological barriers. In these days of ORM many programmers mistakenly believe their class files define reality, but this is not true. Reality is defined by the users who one way or another create the paychecks, and by the database, which is the permanent record of facts. But a programmer who thinks his classes define reality simply cannot see this and will reject the trigger solution for any number of invalid reasons.

So denormalizing by putting in derived values can make a database much more valuable, but it does require a clear systematic approach to generating the derived values. There is no technical problem associated with ensuring the values are correct because of course the application has to do that somehow somewhere anyway, the real barriers tend to be the psychological and political.

The second set of arguments for denormalization tend to be rather weak, and come down to something like this (you have to picture the programmer whining like a child when he says this), "I don't like my data scattered around so many tables, can't we play some other game instead?"

Many programmers, when they first learn about normalization and build a normalized database, discover that the data they need to build a screen is "scattered" about in many tables, and that it is tedious and troublesome to get it all together for presentation to the user. A simple example might be a contacts list. The main table is CONTACTS, and it contains not much more than first and last name. A second table is a list of PHONES for each contact, and a third table is a list of various mailing addresses. A fourth table of EMAILS stores their email addresses. This makes four tables just to store a simple contact! We programmers look at this and something inside of us says, "That's just way too complicated, can't I do something else instead?"

This is a case of programmer convenience clashing with correctness of data. Nobody argues (at least not that I've heard) that they do not want the data to be correct, they just wonder if it is possible to simplify the tables so that they do not have to go out to so many places to get what they need.

In this case, programmers argue that denormalization will make for simpler code if they deliberately skip one or more steps in the normalizing process. (Technically I like to call the result a "non-normalized" database instead of denormalized, but most people call it denormalized, so we will go with that.)

The argument goes something like this: I know for a fact that nobody in the contacts list will have more than 3 emails, so I'm going to skip the EMAILS table and just put columns EMAIL1, EMAIL2, and EMAIL3 into the main CONTACTS table. In this case, the programmer has decided to skip 1st Normal Form and put a repeating group into the CONTACTS table. This he argues makes for simpler database retrieval and easier coding.

The result is painfully predictable. The simplification the programmer sought at one stage becomes a raft of complications later on. Here is an example that will appear trivial but really gets to the heart of the matter. How do you count how many emails a user has? A simple SELECT COUNT(*)...GROUP BY CONTACT that would have worked before now requires more complicated SQL. But isn't this trivial? Is it really that bad? Well, if all you are coding is a CONTACTS list probably not, but if you are doing a real application with hundreds of tables and this "convenience" has been put out there in dozens of cases, than it becomes a detail that programmers need to know on a table-by-table basis, it is an exception to how things ought to be that has to be accounted for by anybody who touches the table. In any shop with more than 5 programmers, whatever convenience the original programmer gained is lost quickly in the need to document and communicate these exceptions. And this is only a single trivial example.

Other examples come when it turns out you need more than three slots for phone. In the normalized case this never comes up. Any user can have any number of phones, and the code to display the phones is running through a loop, so it does not need to be modified for the case of 1 phone, 2 phones, etc. But in the "convenient" denormalized case you now must modify the table structure and the code that displays the contacts, making it quite inconvenient.

Then you have the case of how to define unused slots. If the user has only one email, do we make EMAIL2 and EMAIL3 empty or NULL? This may also seem like a silly point until you've sat through a flamewar at the whiteboard and discovered just how passionate some people are about NULL values. Avoiding that argument can save your shop a lot of wasted time.

In short, programmer convenience should never lead to a shortcut in skipping normalization steps because it introduces far more complications than it can ever pay for.

This blog has two tables of contents, the Topical Table of Contents and the list of Database Skills.

The normalization essays on this blog are:

• Revisiting Normalization and Denormalization (this essay).
• Pay Me Now Or Pay Me Later
• The Argument for Normalization
• First Normal Form
• Second Normal Form
• Third Normal Form and Calculated Values
• The Argument for Denormalization
• Denormalization Patterns
• Keeping Denormalized Values Correct
• Triggers, Encapsulation and Composition
• The Data Dictionary and Calculations, Part 1
• The Data Dictionary and Calculations, Part 2

This is the Database Programmer blog, for anybody who wants practical advice on database use.

There are links to other essays at the bottom of this post.

This blog has two tables of contents, the Topical Table of Contents and the list of Database Skills.

We programmers love to plan for things that will hardly ever happen, like coding the system's upgrade engine to handle spontaneous human combustion, making sure the SQL scrubbing layer can also launch a rocket into space, and, well, trying to work out ahead of time what to do if we ever need a head transplant.

The boss comes over and says, "can you toss a simple plot onto the sales' staff home page that shows sales by day? Use that 'jquicky' or whatever you call it. Should take a couple of hours, right?" And three days later we're working on the world's greatest plotting system that can report everything except what the boss actually asked for because we haven't gotten around to that part of it yet. (Really, can he expect me to just bang this out without the required Seven Holy Layers of Abstraction and Five Ritual Forms of Parameterization, and the Just and Wholesome Mobile Support, or the features Not Yet Required but visible to the Far Seeing Wise and Learned Men?)

So what I am getting at is that programmers of all stripes are addicted to abstraction, it gives us goosebumps and makes us feel warm and tingly, and so we do it even when we do not need to. We build abstraction contraptions.

When it comes to designing a database, this unhealthy proclivity can seriously slow you down, because of what I call:

Everybody wants to be remembered for something. If I could write my own epitaph, it might be:

Table-based datastores are optimally abstract

This law is not about database access, it is about database design. It can be expressed informally as:

People Understand Tables Just Fine

Or more rigorously as:

Table-based datastores are optimally abstract; they require no additional abstraction when requirements are converted to desgin; they cannot be reduced to a less abstract form.

I should point out that this essay deals with structured atomic values, who live in the Kingdom of The Relational Database. The concepts discussed here do not apply to free-text documents or images, or sound files, or any other media.

My basic claim here is that you cannot create an abstraction of data schemas that will pay for itself. At best you will create a description of a database where everything has been given a different name, where tables have been designated 'jingdabs' and columns have been designated 'floopies' and in the end all of your jingdab floopies will become columns in tables. Oh, and I suppose I should mention the Kwengars will be foreign keys and the Slerzies will be primary keys.

After that it goes downhill, because if we generate an abstraction that is not a simple one-to-one mapping, we actually obscure the end goal. Consider an example so simple as to border on the trivial or absured. Why would we ever use the terms "One-to-Many" or "Many-to-Many" when the more precise terms "child table" and "cross-reference table" convey the same idea without the noise? I said above that this would sound trivial, and you can accuse me of nit-picking, but this is one of those camel's nose things, or perhaps a slippery slope. When technical folk get together to design a system, we should call things what they are, and not make up new words to make ourselves sound smarter.

The second half of Ken's law says that you cannot de-Abstract a table schema into some simpler form. This should be very easy to understand, because relational databases deal with atomic values, that is, values which cannot themselves be decomposed. If you cannot decompose something, then it cannot be an abstraction of something more specific.

Going further, if the schema has been normalized, then every fact is stored in exactly one place, so no further simplification is possible. If you cannot simplify it or resolve it into something more specific, then it is not an abstraction of something else.

I originally began to suspect the existence of what I call in all humility "Ken's Law" when I was sitting at a large conference table with my boss, her boss, a couple of peers, and 3 or 4 reps from a Fortune 5 company. My job was basically to be C3PO, human-cyborg relations. Some people at the table protested loudly to being non-technical, while others had technical roles. But everybody at the table spent all day discussing table design.

Later on, when at a different position, the programmers received their instructions from Project Managers. The best Project Managers worked with their customers to figure out what they were trying to keep track of, and handed us specs that were basically table layouts. The customers loved these guys because they felt they could "understand what the project manager was talking about", and the project managers, who swore they were not technical, were respected because they handed us requirements we could actually understand and implement.

Since that time I have further learned that it is very easy for anybody who deals with non-technical people to bring them directly to table design without telling them you are doing it. All you have to do is listen to what words they use and adopt your conversation accordingly. If they say things like "I need a screen that shows me orders by customer types" they have told you there will be a table of customer types. Talk to them in terms of screens. If they say, "Our catalog has 3 different price list and four discount schemes" then you know that there will be a PRICELIST table, a DISCOUNTS table, and likely some cross-references and parent-child relationships going on here.

One of the greatest abstraction contraptions of this century (so far), is ORM, or Object-Relational Mapping, which I do not use precisely because it is an abstraction contraption.

To be clear, the mistake that ORM makes is not at the design phase, but at the access phase. The ORM meme complex instructs its victims that it is ok to put structured atomic values into a Relational Database, but when it comes time to access and use that data we will pretend we did not put it into tables and we will pretend that the data is in objects. In this sense the term Object- Relational Mapping is a complete misnomer, because the point is not to map data to objects but to create the illusion that the tables do not even exist.

Perhaps ORM should stand for Obscuring Reality Machine.

So what about that weird title involving head transplants? Obviously a head transplant is impossible, making it also very unlikely, besides being silly and non-sensical. It came to mind as a kind of aggregrate of all of the bizarre and unrealistic ideas about abstracting data designs that I have heard over the years.

One of these ideas is that it is possible and beneficial to create a design that is abstract so that it can be implemented in any model: relational, hierarchical, or network. I'm not saying such a thing is impossible, it is likely just a small matter of programming, but for heaven's sake, what's the point?

So don't waste time creating abstractions that add steps, possibly obscure the goal, and add no value. Don't plan for things that are not likely to happen, and avoid abstraction contraptions.

This blog has two tables of contents, the Topical Table of Contents and the list of Database Skills.

Other philosophy essays are:

• Prepare Now For Possible Future Head Transplant (This Essay)
• The Quest for The Absolute
• I Am But A Humble Filing Clerk
• Why I Do Not Use ORM
• Minimize Code, Maximize Data

It seems strange to me that I've been working on this blog for 3 years or so (with one very long break) and somehow never got around to writing a simple list of skills that all database experts need. So here it is!

There are three common job titles in the database area, which are Database Administrator (DBA), Database Programmer, and Database Architect. These titles tend to be somewhat variable from shop-to-shop, but generally the "Architect" term indicates the highest level of skill combined with considerable management responsibilities. The "Programmer" term is somewhere below that, but the "DBA" is extremely variable. I have seen shops where a person was called a DBA and filled a relatively constrained role closer to IT or operations (routine tasks, no real programming) and other shops where a person with the DBA title was basically the Architect.

Because of this high variability in what titles mean, I am not going to waste time categorizing skills as belonging to one job title or another, I am simply going to list them all out.

The various levels of skills are these:

• Before Hello World!: The basics of tables, columns, rows
• The Hello World! Level: SQL Select
• Just after Hello World!: Writing Data
• Commands to create, modify and drop tables, or Data Definition Language (DDL)
• Knowing how to use a Query Analyzer or optimization tool
• Understanding Normalization
• Understanding Denormalization
• Understanding Primary Keys, Foreign Keys and Constraints
• Understanding Transactions
• Understanding ACID
• Understanding Indexes as optimization tool
• Views
• Database Security
• Upgrades and Installs
• Efficient access of database from application
• Bulk operations: loading or exporting large amounts of data
• Understanding of Contexts and how they lead to different sets of Best Practices
• Preventing performance degradation through various maintenance tasks
• Deployment strategies: partitioning, tablespaces
• Deployment strategies, failure protection, from simple backup to hot standbys
• Server side coding: stored procedures and functions
• Server side coding: triggers
• Temporary tables

As long as that list is, it only covers those of us who use database systems. There is an entire set of skills for those who actually create and maintain these systems, but that is not something that will be treated in this blog.

If you have never so much as typed a single SQL command, or seen a table diagram, or anything like that, then it is worth a few minutes to go through the basics of what a database does, which is to organize atomic values into tables.

I am going to write an essay on this soon, even though it may seem so obvious as to be completely unnecessary. But I will do it because the most popular essay on this blog is about using GROUP BY, which tells me newer programmers are starving for useful tutorials at the beginner level. So it seems to me, why not put something out there at the very beginning of the beginning?

If you are starting completely from scratch and want to know about database programming, you want to start with the SQL SELECT command. This is the (almost) only command used to extract data from a database, and all of the possible ways to combine, filter and extract data are expressed in the many clauses of this command.

• Simplest Possible SQL SELECT commands
• Simple embellishments: renaming columns, calculations
• Aggregrations: GROUP BY...HAVING
• Multiple table queries: UNION and JOIN (part 1) and JOIN (part 2)
• Subqueries as column values
• Subqueries as tables
• Partitioning Functions: ntile(), row_number(), etc.
• Recursive queries using Common Table Expressions
• Extracting results as XML

When it comes time to change the data in a database there are three commands, listed below. These commands are based on the tables-and-rows nature of databases, and allow to add a row (or rows), change a row (or rows) and delete a row (or rows).

• The INSERT command
• The UPDATE command
• The DELETE command

The term "DDL" stands for "Data Definition Language" and includes all of the commands use to build the tables that will hold the data for the INSERT, UPDATE, DELETE and SELECT statements. The basic list of commands to be familiar with is:

• Understanding Data Types (databases are strongly typed)
• CREATE TABLE and ALTER TABLE
• Commands to add and drop primary keys
• Commands to add and drop foreign keys
• Commands to add and drop constraints
• Commands to add and drop indexes

There are also numerous commands that are specific to different products. Those will not be listed here today, but who knows what the future may bring.

Database programmers, once they get started with the skills listed above, tend to become more and more obsessed with performance. Every major database has some type of tool that lets you examine how the server is going to process a SQL SELECT, and database programmers depend on these tools to discover where they might alter tables or indexes or the SELECT itself to make the queries go faster.

The term "normalization" refers to the process of analyzing the data that your system is required to store, and organizing it so that every fact is stored in exactly one place. Understanding how to normalize data is an absolute requirement for the database programmer who wants to design databases.

We speak of normalization in "forms" as in "first normal form", "second normal form", and so on. It is a good idea to understand The argument for normalization and then to pursue at very least:

• First Normal Form
• Second Normal Form
• Third Normal Form
• Fourth Normal Form and higher forms

Normalization is a a fascinating topic to study, and it extends all they way up to "Domain-key Normal Form" which is considered the most complete normalization for a database.

Every database programmer, after fully understanding normalization, realizes that there are severe practical problems with a fully normalized database, such a database solves many problems but generates problems of its own. This has led programmer after programmer down the path of denormalization, the deliberate re-intoduction of redundant values to improve the usability of the database.

There is a surprising lack of material available on the web regarding denormalization strategies. Most of what you find is arguments and flame wars about whether or not to do it, with little to nothing on how to actually do it. For this reason, I provide my own essays on this blog on the strategies and methods I have worked out over the years:

After reviewing The Argument For Denormalization it is worthwhile to follow up with:

• Understanding the "Automation Constraint" and how to Keep denormalized values correct
• Understanding the three denormalization patterns of FETCH, EXTEND, and AGGREGATE
• Other Patterns

The arguments for and against denormalization are heavily affected by the Pay me now or pay me later design tradeoff.

One might argue that this list of skills belongs much higher up the list, up there with the CREATE TABLE command. However, I have it useful to distinguish between simply knowing the commands to make a primary key and actually understanding the tremendous power of keys.

In this author's opinion it is not truly possible to understand how powerful and beneficial Primary keys and Foreign Keys are for an entire application stack until you have learned the commands, built some databases, and worked through the concepts of normalization and denormalization. Only then can you revisit these humble tools and realize how powerful they are.

• Primary Keys and Table Design
• Choosing data types for primary keys
• Foreign keys and table design
• Foreign keys and cascading actions
• The How and Why of Constraints

The word "transaction" has two meanings in common day-to-day database talk. One meaning is very loose and refers to some individual command or set of commands. You might hear somebody using the term loosely when they say, "We're seeing about 10 transactions per second this week."

The more rigorous use of the term refers to a statement or set of statements that must be guaranteed to either complete in their entirety or fail in their entirety. This is a profoundly important concept once you get beyond simply making tables with keys and get into real-world heavy multi-user activity. And this leads us to the next topic...

Modern relational databases expect multiple simultaneous users to be writing and reading data all of the time. The term "ACID Compliance" refers to both the philosophy of how to handle this and the actual methods that implement that philosophy. The term ACID refers to:

• The Atomic nature of each transaction
• The Consistentcy of the database during and after simultaneous overlapping transactions
• The Isolation of each transaction
• The Durability of the results

An index is a special tool databases use to provide very rapid access to large amounts of data. Just like keys, it is not enough to know the commands, it is necessary to understand the subtle power of indexes when used with some craftsmanship. The basic uses of indexes are:

• A simple index on a column to provide rapid search on that column
• A "covering index" that includes extra columns that can further speed up certain common access patterns
• Clustered indexes (MS SQL Server) and what they give and what they take away
• The cost of indexes on write operations

A view looks like a table to the SQL SELECT command. The view itself is a stored SQL SELECT command that encodes some query that is either used very often or is very compex. In all cases, views are used to present the database data to the application in some simplified convenient or secure form. The two major uses of views are:

• To simplify the application programmer's job
• To provide a read-only interface for some applications

If you are a single programmer or hobbyist working with a database, it is all well and good to just add and drop tables as you wish. But as soon as you get into development with quality control stages and multiple programmers, it becomes evident that you need a strategy for handling the schema changes that come with with new versions of the system. There are multiple essays available on this blog, covering:

• Dictionary based upgrades
• Upgrading indexes and keys
• Upgrades, dictionary, and calculated values part 1
• Upgrades, dictionary, and calculated values part 2

Databases provide incredible security provisions that are just about completely ignored by modern web developers. Sometimes there is good reason for this, but overall anybody who wants to become a truly accomplished Database Programmer or Database Architect must have a thorough understanding of database security and how it can simplify the entire system stack.

Database security comes down to specifying who is allowed to perform the 4 basic operations of INSERT, UPDATE, DELETE and SELECT against which tables:

My basic introduction to security is here.

• Understanding roles (we used to say users and groups)
• Simple table-level security
• Column-level security (not widely supported)
• Row-level security (not widely supported)

Imagine you have the perfectly designed database, with every nuance and subtlety excellently crafted in the ares of keys, indexes, normalization, denormalization and security. At this point your job branches out into several new areas, but one of the most important is knowing how to write application code that efficiently accesses the database.

Some database applications involve a large number of small transactions, where each trip to the database writes only a single row or reads only a dozen or so rows.

But in many cases you need to bulk load large amounts of data in one shot, thousands or even millions of rows. In these cases the techniques that govern small transactions are useless and counter-productive, and you need to learn some new commands and strategies to handle the bulk loads.

Not all databases are created for the same purpose. If you have a very large operations then it will likely have multiple independent databases that fill the classical roles, while in a smaller shop the roles may be combined in one database. I like to refer to these roles as "contexts" because they determine how the tables will be designed and how acess to the tables will be governed. The major contexts are:

• OLTP or Online Transaction Processing, characterized by simultaneous reads and writes, generally assumes little or no periods of inactivity, and generally assumes that the individual transactions are very small. The apps we were all writing in the 80s and 90s to do accounting, ERP, MRP, Job control, payroll, airline reservations and many others fall into this context.
• Data Warehouse context, characterized by periodic bulk loads of new information with most activity being reads. The Data Warehouse context is largely associated with the "Star Schema" table design. Data in a Warehouse is historical, it never changes after it is loaded.
• CMS or Content Management System, also characterized by very few writes compared to reads, but more likely to have a normalized structure. Unlike a Data Warehouse, the data is subject to change, just not that often.
• Any other Read Only Context. I include this category because I spent some time working on Direct Marketing databases, which are like a Data Warehouse in that they are updated periodically and the data does not change, but the Star Schema is completely inappropriate for them.

If you consider a huge online shopping system, you can see that within that application there are at least two contexts. The product catalog is likely to see vastly fewer writes than reads, but the shopping cart tables will be in a constant state of reads and writes.

Once the database and its application stack is up and running, and the reads and writes and coming through, the laws of thermodynamics come into play and system performance can begin to degrade even if the database stays the same size and the load on the system is steady.

Different vendors have different tools for combatting this, but they tend to come down to reclaiming temporary space and occassionally rebuilding indexes. There are also log files that have to be purged, regular backups to be made, and other operations along those lines.

When systems become sufficiently large, it is no longer possible to just attach some disks to a box and run a database server. The Database Architect must consider breaking different tables out onto different sets of spindles, which is usually done with "tablespaces", and moving older data onto slower cheaper spindles, which is often done with Partitioning.

Because a database typically experiences simultaneous reads and writes from multiple sources, and may be expected to be up and running 24/7 indefinitely, the concept of making a backup and recovering from a failure becomes more complicated than simply copying a few files to a safe location.

In the most demanding case, you will need to provide a second complete box that can become fully live within seconds of a disastrous failing of the main box. This is called various things, but Postgres calls it a "hot standby" in version 9 and some MS SQL Server shops call it a "failover cluster."

The ability to come up live on a second box when the first one fails is made possible by the way databases handle ACID compliance, and the fact that they produce something called a Write-Ahead-Log (WAL) that can be fed into a second box that "replays" the log so that its copy of the database is getting the same changes as the master copy.

I really could not figure out where to put this entry in the list, so I just punted and put it near the end. It could really go anywhere.

Stored procedures or functions are procedural routines (not object oriented) that are on the database server and can be invoked directly from an application or embedded inside of SQL commands. Generally speaking they provide various flow-control statements and rudimentary variable support so that you can code multi-step processes on the server itself instead of putting them in application code.

Triggers are quite possibly the most elegant and beautiful technology that servers support, absolutely the least understood, and definitely the most reviled by the ignorant. You will find virtually no web content today that will explain why and how to use triggers and what they are good for.

Except of course for my own essay on triggers that discusses them in terms of encapsulation. Temporary tables

Temporary tables are like Stored Procedures inasmuch as I had no idea where to put them in the list, so they just ended up at the end.

As the name implies, a temporary table is a table that you can create on-the-fly, and which usually disappears when your transaction is complete. They are most often found in Stored Procedures. They can impact performance for the worst in many ways, but can be extremely useful when you are doing multi-staged analsysis of data in a Data Warehouse (that's where I use them the most).

This week The Database Programmer returns after almost 18 months with an entry on using Common Table Expressions (CTEs) to do recursive queries. Relational databases were plagued from their inception with a lack of meaningful treatment for recursive operations, and CTEs have finally plugged that hole.

Common Table Expressions appeared in SQL Server 2005, and in PostgreSQL 8.4, and are also available in Oracle. As for mySQL, since I don't use it, I did a quick Google search and looked at the Docs for 5.5, and couldn't really find anything. I generally tend to assume mySQL cannot do the tough stuff.

There have always been plenty of people who claimed SQL was a bloated and clumsy language to work with. Most of the time I tend to agree, but I find the advantages of relational/SQL system to be so large that I'm willing to pay that price.

But with Commom Table Expressions (CTEs) I just can't help drifting into conspiracy theories involving the enemies of SQL infiltrating the committees and deliberately suggesting the most twisted, bloated, complicated way they could think of to do what is really a very basic operation. In other words, I am profoundly unimpressed with the syntax of CTEs, but as long as they are here and they work, we'll go along.

Your basic recursive table contains a foreign key to itself, so that some rows in the table are children of some other row in the table. This recursion can nest to any depth, and the chart below shows a very simple example:

Primary_key   |   Parent_Key  |  Notes  
--------------+---------------+---------------------------
     A        |     null      |   top level row, no parent
     B        |      A        |   first level child of A
     C        |      B        |   child of B, grandchild
              |               |   of A
     D        |      C        |   child of C, grandchild 
              |               |   of B, great-grandchild 
              |               |   of A
     X        |     null      |   top level row, no parent
     Y        |      X        |   child of X
     Z        |      Y        |   child of Y, grandchild 
              |               |   of X

What we want is a query that can return a given row and all of its children out to any level, including helpful information about the structure of the recursion, something like this:

Primary_key | Level | Top_Key | Immediate_Parent_key 
------------+-------+---------+-----------------------
     A      |   1   |  A      | null
     B      |   2   |  A      | A    
     C      |   3   |  A      | B   
     D      |   4   |  A      | C
     X      |   1   |  X      | null
     Y      |   2   |  X      | X   
     Z      |   3   |  X      | Y   

At this point the mind boggles at how long this blog entry needs to be to explain this simple operation. But lets get going anyway.

A Common Table Expression begins with the "WITH" clause and ends with a standard SQL Select:

;WITH myCTEName (primary_key,level,top_key,immediate_parent_key)
as (
  ....we'll get to this below....
)
select * from myCTEName

The basic idea is that we are going to define a CTE with a name and a list of columns, and then SELECT out of it. Without that final SELECT statement the CTE does not actually do anything. The SELECT can also be arbitrarily complex, doing aggregrations, WHERE clauses and anything else you might need.

The first thing to notice is the leading semi-colon. This is a trick adopted by MS SQL Server users. SQL Server does not require statements to be terminated with a semi-colon, but a SQL Server CTE requires the previous statement to have been terminated with a semi-colon (nice huh?). So SQL Server programmers adopted the strategy of starting the CTE with a semi-colon, which keeps the syntactical requirement with the CTE, where it belongs.

A given CTE sort of has a name. That is, you have to name it something, but think of it as a table alias in a SQL SELECT, such as "Select * from myTable a JOIN otherTable b...", it exists only during the execution of the statement.

The columns listed in the parantheses can have any names (at least in SQL Server). But these column names are what you will refer to in the final SQL SELECT statement.

Now we code the inside of the CTE in two steps. The first step is called the "anchor", and it is a straightforward query to find the top-level rows:

;WITH myCTEName (primary_key,level,top_key,immediate_parent_key)
as (
    select primary_key   as primary_key
         , 1             as level
         , primary_key   as top_key
         , null          as immediate_parent_key
      from myRecursiveTable
     where Parent_key is null
)
select * from myCTEName

This should be self-explanatory, we are querying only for rows that have no parent (WHERE Parent_key is null) and we are hardcoding the "level" column to 1, and we are also hardcoding the "immediate_parent_key" column to null.

This query alone would return two of the rows from our desired output:

Primary_key | Level | Top_Key | Immediate_Parent_key 
------------+-------+---------+-----------------------
     A      |   1   |  A      | null
     X      |   1   |  X      | null

Now we are going to add the actual recursion. When I first learned CTEs this was the hardest part to figure out, because it turned out my hard-won set-oriented thinking was actually slowing me down, I had to think like a procedural programmer when defining the second half of the query.

;WITH myCTEName (primary_key,level,top_key,immediate_parent_key)
as (
    select primary_key,1,primary_key,null
      from myRecursiveTable
     where Parent_key is null
    UNION ALL
    select chd.primary_key,par.level+1,par.top_key,chd.parent_key
      FROM myCTEName        par
      JOIN myRecursiveTable chd ON chd.parent_key = par.primary_key
)
select * from myCTEName

Thinking step-wise, here is what is going on under the hood:

1. The server executes the "anchor" query, generating a result set called "myCTEName" containing just the top level rows.
2. The server then executes the second query. At this point the result set "myCTEName" exists and can be referenced, so that you can link children to their parents. (That's why you see the JOIN)
3. Step 2 is repeated recursively, adding grand-children, great-grand-children, and so on, until no more rows are being added, at which point it stops, and...
4. The final result set is passed to the trailing SELECT, which pulls results out of "myCTEName" as if it were a table or view.

So when we code the 2nd part of the inside of the CTE, the part after the UNION ALL, act as if the first query has already run and produced a table/view called "myCTEName" that can be referenced. Once you understand that, the query is pretty easy to understand:

• The "From myCTEName par" clause tells us we are pulling from the previously generated set. I like to use the alias "par" for "parent" to remind myself that the prior result is the parent row.
• We then join to the original source table and use the alias "chd" to remind ourselves we are pulling child rows from there. The "ON chd.parent_key = par.primary_key" defines how children are joined to parents.
• Our first column, "chd.primary_key", is the unique key for the results.
• Our second column, "par.level+1" gives us a nifty automatically incremented "level" column.
• Our third column, "par.top_key" ensures that all rows contain a reference to their top-most parent.
• Our final column, "chd.parent_key", makes sure each row contains a reference to its immediate parent.

Once you have the inside of the CTE coded, the fun part moves to the final SELECT, which is operating on the complete set of results. You do not necessarily have to pull the complete list. For instance, you may want to find out the maximum nesting level for each parent, or the count of children for each parent:

;WITH myCTEName (primary_key,level,top_key,immediate_parent_key)
as (
    select primary_key,1,primary_key,null
      from myRecursiveTable
     where Parent_key is null
    UNION ALL
    select chd.primary_key,par.level+1,par.top_key,chd.parent_key
      FROM myCTEName        par
      JOIN myRecursiveTable chd ON chd.parent_key = par.primary_key
)
select top_key
     , max(level) as maxNestingLevel
     , count(*)   as countRows
     , count(*)-1 as countChildren
  from myCTEName

Common Table Expressions give SQL-based databases the (very) long-needed ability to execute recursive queries, albeit with a rather complex syntax. Once you grasp the basics of how to code them, there are many possible uses that go far beyond the simple example shown here.