update statistics concurrency with application SQL

Is there any real or serious problem with running update statistics concurrent with application's SQL ? Does locking occur that severely impacts contention between update statistics and application's SQL ? Is there any known case/experience of table corruption due to running update statistics while application's SQL are running concurrently ?Are there specific configuration setting to check or revise in order to run update statistics and application's SQL concurrently ? We make extensive use of referential integrity - is/should this be a factor in any way ? Does anyone have comparisons of duration to run update statistics by itself versus running it under a light, medium and heavy work load ? We are running a 56GB instance using IDS 7.30UC8, HPUX 11.0. We run update statistics in database exclusive mode every two weeks on about150 tables. We first drop all distributions table by table and re-create the distributions as Informix recommends. I appreciate your help. Sent via Deja.com http://www.deja.com/ Before you buy.

In article <85ge56$se5$1@nnrp1.deja.com>, roswil@my-deja.com wrote: > Is there any real or serious problem with running update statistics > concurrent with application's SQL ? Does locking occur that severely > impacts contention between update statistics and application's SQL ? > Is there any known case/experience of table corruption due to running > update statistics while application's SQL are running concurrently ?> Are there specific configuration setting to check or revise in order to > run update statistics and application's SQL concurrently ? We make > extensive use of referential integrity - is/should this be a factor in > any way ? > Does anyone have comparisons of duration to run update statistics by > itself versus running it under a light, medium and heavy work load ? > > We are running a 56GB instance using IDS 7.30UC8, HPUX 11.0. We run > update statistics in database exclusive mode every two weeks on about> 150 tables. We first drop all distributions table by table and > re-create the distributions as Informix recommends. > I appreciate your help. > Anyone ? thanks Sent via Deja.com http://www.deja.com/ Before you buy.

roswil@my-deja.com wrote: > > In article <85ge56$se5$1@nnrp1.deja.com>, > roswil@my-deja.com wrote: > > Is there any real or serious problem with running update statistics > > concurrent with application's SQL ? Does locking occur that severely > > impacts contention between update statistics and application's SQL ? > > Is there any known case/experience of table corruption due to running > > update statistics while application's SQL are running concurrently ?> > Are there specific configuration setting to check or revise in order > to > > run update statistics and application's SQL concurrently ? We make > > extensive use of referential integrity - is/should this be a factor in > > any way ? > > Does anyone have comparisons of duration to run update statistics by > > itself versus running it under a light, medium and heavy work load ? > > > > We are running a 56GB instance using IDS 7.30UC8, HPUX 11.0. We run > > update statistics in database exclusive mode every two weeks on about> > 150 tables. We first drop all distributions table by table and > > re-create the distributions as Informix recommends. > > I appreciate your help. There will be a performance impact. The only locking will be an instantaneous lock on the tables' records in the systables, sysindexes, and sysdistrib tables unlikely to cause any concurrency problems. There IS a possibility that the new stats will invalidate any prepared statements'/cursors' query plan causing a -710 error. Informix says they are trying to eliminate that problem but for now your apps may have to reprepare cursors if they get a -710 caused by the update stats. -- Art S. Kagel & Family kagel@erols.com