Re: SMP and 7.x performance questions?

This is a multi-part message in MIME format. --------------675362CD5454 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Joe Lumbley wrote: > Michael Mstowski (msm@spaceworks.com) wrote: > : I talked to some db programmers and they mentioned that > : Informix 7.x has poor performance running on a machine > : with 1 or 2 CPUs. It was mentioned that I would be better > : off not upgrading (FOR PERFORMANCE), unless I had at least > : 4 CPUs. Can anyone verify or clearify the above statement? > > That's what I had heard also, and it's wrong. Informix 7.X should > give you **better** performance than earlier versions even on > non-multi-processor systems, due to the improved architecture. Well, I'm afraid this ain't necessarily so. For *most* types of applications, 7.X will run faster. For example, complex DSS-type queries will generally run faster under 7.X, even on a 2 cpu box. On the other hand, OLTP type applications may see some degradation. At least part of this is due to an increased code path to achieve the same results; the complexity of the server increased substantially from 5.X to 7.X, and there are just a few more layers to work with for even the simplest queries. The most obvious impact comes with OLTP type queries run with a small (>10) number of users. In these cases, 5.X may well run faster. As the complexity of the query grows, and as the simultaneous user count grows, the curve comes back quickly and 7.X tends to outperform the eralier versions. A small SMP is a bit problemmatic, though. For some reason, it gets hard to load balance across only 2 cpus. For those in such a situation, I recommend trying to tun the system as though it were a uniprocessor (set the SINGLE_CPU parameter on). This introduces some optimizations through the code stream (avoids doing things that are only pertinant in a multi-proc environment) that can help. Once you are 4 procs or better, though, you are better off tuning for SMP. > 7.X is a little bit harder to tune though. This is because of the > additional options and parameters that it supports. The onstat utility > is ***greatly*** improved, although somewhat sparsely documented. Much of the onstat output that is undocumented is meaningless without an understanding of complex internals. This is not to say that all the undoc'ed stuff requires a rocket scientist; just that some of it isn't worth bothering to document for the vast majority of users. -- Dave Kosenko, Informix Professional Services ****************************************************************************There's more than one way to skin a cat; the cat himself, however, generally fails to appreciate the differences. --------------675362CD5454 Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit Content-Disposition: inline; filename="DISCLAIM" ************************************************************************* Disclaimer: All opinions expressed in this message are well-reasoned and insightful; needless to say, they are not those of Informix Software, its partners or lackeys. Anyone who says otherwise is itching for a fight. --------------675362CD5454--