Re: Performance advice

Andrew Hamm wrote: > > Richard Kofler wrote: > > I have a question in this context: > > > > We have 32 CPUs an run 28 CPVPs. (Solaris 7 , IDS V9.40.UC3W1) > > > > All connections come in via a serverside application bundling db > > requests > > for approx 3k users, partly still running on the database server: > > (but we wil be able to put that app on a different hardware very soon) > > 70 connection via ipc (from the very same machine the database > > instance > > runs on), 70 via TLI (from an other piece of hardware - 1Gbit > > connected), > > both groups of connections with a real lot of traffic. > > So you are saying that the 3k users go via a small number of connections no. The application has proxies and a transaction manager is part of what I call app here. Finally the IFX 'clients' are 2 times 70 unix processes running on 2 boxes So it is 2 times 70 sessions and some onstat or onbar or other utils > (either independent or part of the 70+70 connections) plus the 70 + 70 > connections? > > Therefore your connections would be approx 150 to 200? > > If so, then that's not really enough connections to actually demand so much > load sharing from the engine. If all 3000 users were connecting > independently, then you'd see the phenomenum (help captain) we are talking > about. > > Or, maybe you've got TOO MANY CPU's. (Is that possible???) no, not too many CPUs. After tuning I/O over the top we are in fact running CPU bound thanks to not so very optimal index based filtering manly done incore (i.e. in the LRU buffers) I mean, we do *not* have the slightest I/O bottleneck anymore, but that is where we came from, 4 weeks ago. At least we are now able to do somehing like 90 TX / second but we sort of need like 120 / sec to make cutomer happy I mayhaps should also mention, that we really have a nice nasty collection of lil devils here: ANSI type logging, XA transactions and other stuff usually found mainly in DBA nightmares :) dic_k -- Richard Kofler SOLID STATE EDV Dienstleistungen GmbH Vienna/Austria/Europe

Richard Kofler wrote: > > no. > The application has proxies and a transaction manager is part > of what I call app here. >[SNIP ETC] anyway, nfi. I've definitely seen the load imbalance once I went looking for it after Art's postings. This is on machines with a small number of CPU's eg 3,4,5 Maybe there's some interesting large-scale factor kicking in for your system.