Re: A performance question

In article <330B7E2B.6916@thetsn.com>, Gregory Trubetskoy <grisha@thetsn.com> writes >I am evaluating Informix on NT (to possibly replace MS SQL Server), and >I notice a significant difference in performance. > >I did a little tweaking, not too much - I set BUFFERS to 6000 (x4096K >page)(the machine has 128 Meg) SHMVIRTSIZE to 16Meg, MULTIPROCESSOR 1 >(the machine is a dual Pentium). > >Anyhow, I have duplicated a table that I have on SQL server. About 4 Mil >rows. I noticed that a SQL statement that I'm trying returns almost >immediately on SQL Server, especially the second time (because of >caching, I guess), but takes significantly longer on Informix. And >output of the same query running the second time isn't any faster. > >The SQL statement is such that it has to create a temporary table (there >is GROUP BY) and has to do some index scanning (there is an "IN (x, x, >x, x, x, x)" clause). > >Both SQL Serv. and Inf. have exactly the same clustered index built, >looking at query plans shows (EXPLAIN ON and SHOWPLAN ON)that they both >use the index in the query - yet Informix is about 15 times slower. I am >positive I don't have somethign configured correctly - if anyone can >understand any of the above and has some advice, I'd appreciate it. > >Thanks. Do alter index to cluster on the informix index and try again. -- David Williams