Hi Holger/Christian, I thought of the cache issue, and therefore ran the command numerous times from SQL Studio. The first time is much longer for the Count(ID) and then it drops to 4.5 seconds consistantly thereafter. The time for count(*) is not much different the first time, and then remains at 0.1 seconds.
How can this be?!!! Cheers, David "Becker, Holger" <[EMAIL PROTECTED]> on 07/10/2003 03:42:28 AM To: "'[EMAIL PROTECTED]'" <[EMAIL PROTECTED]>, [EMAIL PROTECTED] cc: Subject: RE: time difference between count(*) and count(id) [EMAIL PROTECTED] wrote: > Have seen postings before that count(*) is slower than counting on an > index. > > However, I am seeing count(*) perform faster than count(id)!! > > Explain for Count(*) is: > > KIOSK METRICS METRICS_KIOSKID_EXECTIME_COPIES INDEX SCAN > 1206 * > KIOSK RESULT > IS COPIED , COSTVALUE IS 1208 > > and takes 110 milliseconds. > > Explain for count(id) is: > > KIOSK METRICS METRICS_KIOSKID_EXECTIME_COPIES INDEX SCAN > 1209 * > KIOSK RESULT > IS COPIED , COSTVALUE IS 1211 > > and takes approx 4.5 seconds!!! > > Can someone please explain? (no pun intended!) Because of the same strategy in both cases I assume you see caching effects. To check this out you have to measure the execution time of both statement right after the restart of the db. Kind regards, Holger SAP Labs Berlin -- MaxDB Discussion Mailing List For list archives: http://lists.mysql.com/maxdb To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
