RE: query performance SE vs IDS

-----Original Message----- From: owner-informix-list@iiug.org [mailto:owner-informix-list@iiug.org] On Behalf Of Obnoxio The Clown Sent: Wednesday, December 31, 2003 3:35 PM To: informix-list@iiug.org Subject: Re: query performance SE vs IDS Jack wrote: > "Art S. Kagel" <kagel@bloomberg.net> wrote in message > news:<pan.2003.12.30.15.46.11.80429.12806@bloomberg.net>... >> On Tue, 30 Dec 2003 06:19:42 -0500, Jack wrote: > > > > SE 7.25uc4 and IDS 9.30uc1 on some linux 2.4. > i put the rootdbs chunk and se database in the same filesystem. > i used stores7 database and the query file is : > set isolation to dirty read; -- ids > select fname from customer where customer_num=101; > select fname from customer where customer_num=102; > etc... i know quite stupid. > and executed with dbaccess stores7 query.sql. > ids gives very good response time when more than 1 user is working > and with more complex queries. > > OB wrote: >What you're seeing is pretty much par for the course. Another place where IDS will run better is if the tables are really big. But you have to >remember the code path for SE is much shorter than IDS, you'd probably find >on this sort of query with a small database that SE would be faster than >OnLine would be faster than IDS 7.x would be faster that 9.2x or 9.3x. One >key point -- is your filesystem journalled? If it is, try making a raw >partition for IDS and see if that isn't just a tiny bit faster... :o) Interesting point - IDS9 would probably be faster if the table was really big (millions of records ...?) - built for size and complexity ________________________________ << ella for Spam Control >> has removed 280 Spam messages and set aside 0 Newsletters for me You can use it too - and it's FREE! www.ellaforspam.com sending to informix-list

Dirk Moolman wrote: >> OB wrote: > >>What you're seeing is pretty much par for the course. Another place where > IDS will run better is if the tables are really big. But you have to >>remember the code path for SE is much shorter than IDS, you'd probably >>find >>on this sort of query with a small database that SE would be faster than >>OnLine would be faster than IDS 7.x would be faster that 9.2x or 9.3x. One >>key point -- is your filesystem journalled? If it is, try making a raw >>partition for IDS and see if that isn't just a tiny bit faster... :o) > > Interesting point - IDS9 would probably be faster if the table was really > big (millions of records ...?) - built for size and complexity How is this different from "Another place where IDS will run better is if the tables are really big." ? :o) -- "C'est pas parce qu'on n'a rien ''' dire qu'il faut fermer sa gueule" - Coluche