RE: Any exp with NTw/2 to 4 hamster wheels?

(FWIW)
"Conventional NT wisdom" is that NT+Oracle ***LOVES*** multi-cpu.

Similar to my reply to Charlie yesterday, I agree with the theory you are stating. However, the question was in the context of an:

   *observed performance improvement*.

I didn't understand all of it, but Paul Drake posted what looks like a fantastic elucidation of the complexities involved in understanding multi-cpu architectural issues - considering the whole system. Presumably you would want someone with that kind of knowledge to solve configuration planning problems.

On the other hand, w/ NT, you can just go out and buy a couple of extra CPUs for +\- $2,000, and see it if "works better" (however unscientifically that determination is made). If not, take them back out. :)

Cavemen didn't have to know the scientific theory of crystalline molecular structures in order to be able to hit a couple of pieces of rock together to make some damn fine flint blades for hunting. They got the job done anyway. Well, better not start a thread on bar-b-q and hamsters, so i'll stop there.

regards,
ep

On 6 Dec 2000, at 18:20, Henry Poras wrote:

Date sent:      	Wed, 06 Dec 2000 18:20:48 -0800
To:             	Multiple recipients of list ORACLE-L <ORACLE-L_at_fatcity.com>
From:           	Henry Poras <Henry.Poras_at_ctp.com>
Subject:        	RE: Any exp with NTw/2 to 4 hamster wheels?

> There was an interesting article (I believe it was on orapub) dealing with
> this phenomenon. If I remember correctly, the idea was that increasing one
> resource could decrease performance by shifting the existing bottleneck. For
> example, if the instance runs a mix of batch jobs and OLTP, with the batch
> being CPU bound, adding more CPU would allow better batch performance (bbp).
> The batch job might now bottleneck on IO, decreasing OLTP performance.
Received on Thu Dec 07 2000 - 12:45:42 CST