Re: Big question for data 'replication '

Allessandro,

Why do you want to keep the databases aligned? What is the goal to achieve? HA? off-loading queries to a standby? What is the gap allowed? i.e. do you need both db's to be synchronised, or is async replication enough?
If talking HA, what is the failover time allowed, and how much data loss is acceptable?

Because you are using spatial EE is involved already. Data Guard won't cost you extra then (license-wise). The overhead is negligable to very limited too in most cases (for HA I consider physical standby only). How much redo is generated per second/minute/hour at peek times? I'd say the amount of tables is not a parameter in judging the value of a solution. The value of the data to the business is.

What is a DR environment by your definition? Lot's of people think they have HA using RAC: they have n (n>1) db servers talking to 1 database. It protects from server failure, not from db failure. With DG you will have n (n>1) db servers talking to n databases. That is a higher level of HA, IMHO. When set up properly, using a DELAY in applying redo to the standby, you will be protected from human errors (to a certain level/time frame depending how long you configured the delay) as well from hardware errors, albeit not from earthquakes, fires and floods. RAC will include the human errors into the exclusion list as well. I've lots of customers using DG in one building, although they tend to go for a remote standby more and more. But even the 'local' standbys where responsible for nice rescues.

I think (advanced) replication and an application based solution will be more error prone and require lots more of attention/management/maintenance.

There is no straight answer for your question. Investigate risks, requirements ans solutions. Quantify potential losses of risks, and costs of solutions. Do not only include costs of initial setup, but also costs due to (in)flexibility, management, maintenance. Let the business make the final decision. It's their business, it's their data.

Best regards,

Carel-Jan Engel

===
If you think education is expensive, try ignorance. (Derek Bok) ===

On Tue, 2006-08-29 at 12:08 +0200, Alessandro Vercelli wrote:

> Hi all,
> I've inherited a couple of Oracle low-range servers, RedHat ES 3; each node contains an Oracle 10.1 database; both DBs have same physical and structure ans same schemas/tables, the application is installed on a separated machine; the oddity is that the db schema involved in the application contains two (yes, two...) tables of ~300 and ~1.000.000 records, nearly; the schema works with Spatial feature enabled and most probably this has been the requirement for an expensive (and not easy to mantain) Oracle DB.
>
> My question is: assuming I need to maintan both DBs aligned, defining a primary DB and replicating data on the secondary, what are the technical suitable choices ?
>
> On my own, I took into consideration:
> - Data Guard: like driving a Ferrari in a traffic jam, where do you want to go? is it worth such a product for two tables? can we afford the increased resource needs? moreover does DG make sense without a disaster-recovery environment?
> - Replication: it works well enough and not hard to mantain; I know this feature little, is it still implemented? Could you give some hints on how to build it? I used this word in the subject, but it doesn't mean I want to use it;
> - Build on own solution, working at application level or other non-complex DB features (triggers ?); has someone some similar experience to share?
>
> Thanks to everyone who will reply,
>
> Alessandro
>
>
> ___________________________________________________________________
> Salva ora il tuo preventivo gratuito Direct Line e assicurati lo sconto extra 5+5% sulla polizza auto!
> http://click.libero.it/directline2
>
>
> --
> http://www.freelists.org/webpage/oracle-l
>
>

--
http://www.freelists.org/webpage/oracle-l