Re: update statistics not recommended!

Well, given that sharp people have said here that it's true, I'll have to defer to their word on this one, though it still hard to imagine what sort of finagling it takes to produce worse results WITH stats than without. It has been suggested that ERP packages with their middleware layer that does indeed massage the SQL before letting the engine see it is the culprit. FWIW, SAP stongly recommends keeping your Informix stats very current, and I can attest SAP runs like s#it with no stats. How about the other major ERP's, Baan and Peoplesoft, maybe JDEdwards? Can anyone familiar with them post what they recommend??? Greg Bernstein, Rick wrote: > > As strange as it sounds, Informix says to never run Update Statistics > against a pre-IDS 7.30 Lawson database. This is documented in a > presentation given by Tom Reiger (Informix Senior Sales Engineer) > at the 1998 Informix WorldWide User Conference. The presentation > was titled Fully Utilizing IDS with the Major ERP Applications. > He indicated that: > * Lawson queries should always use an index. > * Starting with IDS 7.30, the inxdb7 driver takes advantage > of Informix syntax to address this issue. > He provided SQL syntax to remove statistics from the following tables: > systables, syscolumns, sysindexes, sysdistrib > > > -----Original Message----- > From: jay@ulife.com [mailto:jay@ulife.com] > Sent: Friday, August 14, 1998 18:36 > To: informix-list@iiug.org > Subject: Re: update statistics not recommended! > > Neil Truby <ntruby@netcomuk.co.uk> wrote: > > >My new site has a third-party financial package (Lawsons), running on > >v7.13. The underlying DBMS is Informix. The Lawsons help desk has > told > >me that they recommend that we never run UPDATE STATISTICS, as it will > >degrade performance. Their reasoning, allegedly based on empirical > >evidence of their users, is that all their SQL is written with where > >clauses, and the indexes that exist will be used, which is what is > >required. If we run an UPDATE STATS, we might risk unexpected and > >undesirable access paths. A caveat is added that, under v7.23, some > >customers have experienced benefits from an UPDATE STATISTICS LOW, and > >that we might run it, but prepared to remove the statistics so derived > >if performance nose-dives. > > > >Well, sounds like bollocks to me, but does anyone else have any views > >or, better still, experience of this software packeage running with > >Informix? > > > >Thanks > > > >Neil > > Sounds like some idiot doen't know what they are talking about. You > should run > "update statistics high" for every column that it a lead column in an > index. > > ___________________________________________________________ > Jay Aymond > EXE Technologies > jay_aymond@exe.com