Re: Any Suggestions on Method R

From: Holger Baer <holger.baer_at_science-computing.de>
Date: Thu, 25 Aug 2005 15:30:41 +0200
Message-ID: <dekh62$deb$1@news.BelWue.DE>

Matthias Hoys wrote:

> "Holger Baer" <holger.baer_at_science-computing.de> wrote in message 
> news:deka9m$63s$1_at_news.BelWue.DE...
> 

>>alan_h_r_at_yahoo.com wrote:
>>
>>>Hi,
>>>
>>>I'm a DBA/Developer who has recently learned about the "Method R"
>>>tuning philosophy. So far I've read "Oracle Insights", am reading
>>>"Optimizing Oracle Perfomance" and I have "Oracle Wait Interface"
>>>waiting in my bathroom for it's turn.
>>>
>>>Does anyone who favors "Method R" have any suggestions on where I
>>>should go next? I love the concept though I definitely need more help
>>>with what to do after I target the hogs in my systems.
>>>
>>>Any suggestions?
>>>
>>>Thanks in advance,
>>>
>>>Alan
>>>
>>
>>The problem (if there is any) with "Method R" as Cary Milsap calls it, is
>>that it helps
>>you identify the bottleneck of your system, but of course it can't tell
>>you how to solve
>>that.
>>
>>But having identified the reason for the unsatisfying
>>performance/throughput/whatever,
>>you're pretty much equipped to find a solution, too. But this means mostly
>>talking to
>>
>>* the developers (do you really need to do this like you're doing it),
>>* the managers (this report of yours takes 6 hours to complete, hogging
>>our system no end,
>> but most of the information can be gathered in minutes here)
>>* the vendor (I'm sorry but your software is taking too long in this
>>process, either you
>> fix it or we'll cancel our contract)
>>
>>And a few trips to metalink and several other websites (Asktom, Jonathan
>>Lewis' site, Hotsos
>>to name a few) should get you up to speed.
>>
>>HTH
>>Holger

> 
> 
> My experience is that 60-80% of performance problems are related to 
> inefficient SQL, or to put it in other words : developers/report 

Put that number up to 90% and I'll agree. Real Programming seems to be a lost art what with all those new technologies popping up every 5-10 years. Along with them a new generation of developer arises and they not only don't know a thing about databases, they even pretend they needn't to.

It seems that as soon as Billy runs out of his proverbial lead pipes (or worse, out of coffee), the world as we know it will cease to exist and fall to the hand of the java mob. (Or .Net or what ever will be the new fancy).

> users/application administrators not knowing what they're doing. I once had 
> to "tune" a system where 1 user was constantly consuming 50% of all CPU and 
> 25% of all disk resources because of inefficient SQL usage. Queries that 
> were running for more than 5 minutes (on a OLTP db) were considered "normal" 
> by that person. I know developers with +5 years of Oracle experience who 
> don't know what an explain plan is. My recommendation was that all direct 

We shouldn't call them developers then ;-) (or talk about experience, either).

> end-user access (so not by an application) to the production databases 
> should be restricted as strictly as possible, under direct responsibility of 
> the DBA. But, management often decides differently ...
> 

Cheers,
Holger Received on Thu Aug 25 2005 - 08:30:41 CDT