Re: SeqScans Statistics 7.3_uc6

Hi Art, what is the name of the environment variable you mentionned. TIA, David... ----- Message d'origine ----- De : Art S. Kagel <bloomberg.net!kagel@laposte.atmel.fr> ' : <informix-list@iiug.org> Envoy' : jeudi 25 f'vrier 1999 22:33 Objet : Re: SeqScans Statistics 7.3_uc6 Look for correlated sub-queries. Ver 7.30 unwinds these into joins on the fly. If you have the indexes on all tables involved this produces vastly superior performance to 7.24 for these queries. However, if you do not have the supporting join indexes (ie an index on each table that begins with the join column) then performance is atrocious. There is an environment variable to disable this behavior but the best solution is to add the missing indexes and then update stats again. Let me plug dostats.ec to perform the optimal statistics updates in least time. Art S. Kagel

David Andrieu wrote: > > Hi Art, > > what is the name of the environment variable you mentionned. Set NO_SUBQF in the environment. However, I do not know if it has to be set in the user's environment or the engine's at startup. > > TIA, > > David... > > ----- Message d'origine ----- > De : Art S. Kagel <bloomberg.net!kagel@laposte.atmel.fr> > À : <informix-list@iiug.org> > Envoyé : jeudi 25 février 1999 22:33 > Objet : Re: SeqScans Statistics 7.3_uc6 > > Look for correlated sub-queries. Ver 7.30 unwinds these into joins on > the fly. If you have the indexes on all tables involved this produces > vastly superior performance to 7.24 for these queries. However, if you > do not have the supporting join indexes (ie an index on each table that > begins with the join column) then performance is atrocious. There is > an environment variable to disable this behavior but the best solution > is to add the missing indexes and then update stats again. Let me plug > dostats.ec to perform the optimal statistics updates in least time. > > Art S. Kagel