Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Re: Query for Top 10 Most "Expensive" Queries

Re: Query for Top 10 Most "Expensive" Queries

From: Billy Verreynne <vslabs_at_onwe.co.za>
Date: 28 Aug 2003 22:23:00 -0700
Message-ID: <1a75df45.0308282123.58d5dbe7@posting.google.com>


"Ana C. Dent" <anacedent_at_hotmail.com>

> > Computer resources sitting idle is wasting your money. Computer
> > resources being used is putting your money to use.
>
> The following a a FORTRAN statement
>
> 10 GO TO 10
>
> The statement above is 100% guarenteed to use CPU cycles.
> However, I do NOT consider it to be "putting my money to use".
>
> Never confuse movement with progress.
>
> Going around in circles is movement to the max.

> Most folks will not consider going around in circles to be progress.

I suggest that you read the last paragraph of my posting.

I said: "Thus high resource usage is only expensive when the process has its head up its backside - thus you need to be careful not to judge all all high resource usage queries as expensive. It is more expensive having under utilised resources."

In your example above - that is what I meant with "head up backside".

It is kind of interesting to see people's perspective of today ito what is expensive processing.

In the 80's on mainframes, processing power was in high demand. The workdloads we had often exceeded CPU capacity. We had to schedule workloads in such a way that the mainframes were never idle. Re-configure the mainframes job queues and processing streams for optimal performance. An idle CPU second was more expensive (in real monetary terms) than a used CPU second. By the same token, every used CPU second had to be used *effectively*.

Hell, we had to schedule Cobol compiles and links.. as that was of a lower priority. Depending on the workload that day, you may only get a change to compile your software twice... the result? It only took two compiles, sometimes three, to code software that *works* as spec'ed the first friggen time around. You the programmer made sure that your code required the minimal amount of bug-fixing and modifications to work.. as required.

Today - programmers write crap code. Even after a 100 builds of that software it is still riddled with bugs and does not work per spec. Sysadmins/DBAs get performance concerns when they notice 80% sustained CPU activity. As if a 50% utilisation is more "acceptable". Both groups wet their pants at the first sign of a problem and post messages like URGENT! on Usenet.

It is a sad sight to see.. but also keeps my lead pipe wet and warm.

--
Billy
Received on Fri Aug 29 2003 - 00:23:00 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US