Re: Performance Problem with Temp Table

In article 2iksgmINNlm5@emory.mathcs.emory.edu, warlok!kysilk.sherwood.com!jerry@uunet.UU.NET (jerry) writes: >Well, we have resolved the performance problem with the inserts in the >temp table. > >Thanks to Bob Baskett @ Motorola for the correct and prize winning >answer. The temp table needed to be created "with no log". This >reduced the time from 80 seconds to ~5 seconds. You bet that is going >to become part of my permenant collection! > >Thanks also to Jim Gordon, Paul Hatcher and Troy Hewitt for their >reponses. > >Question to those Informix people who hang out on this group: > >Why such a pronounced difference in the performace of that query with >the "no log" added to the creation of the temp table? Is this >platform/system specific or is it common across all platforms/systems? > > >-- >Jerry M. Denman -- Director of Technical Services 4gen!warlok!jerry >Sherwood Systems (a division of Sherwood Mfg Co. Inc.) jerry@sherwood.com >4760 N. Central Phone:(602) 230-8188 >Phoenix AZ, 85012 FAX:(602) 230-9491 >------------------------------------------------------------------------------ >A true realist demands the impossible > A true idealist demands the impractical Answer to your speed questions - is with no log no regard for the transaction log is created and therefor transaction rollback / commit is truned off so it just flat out loads the data... The time difference - wow that is quite a big difference in time. I could see 2X but you have 16X.