I see the same type of slow downs using 5.0.18 I am using "load data in file" to load CSV files.
with clean tables, I see fairly quick inserts (ie "instant") 2006-08-30 12:07:15 : begin import into table1 2006-08-30 12:07:15: end import into table1 records (10962) >From earlier this morning, before I rotated my tables: 2006-08-30 09:02:01 : begin import into table1 2006-08-30 09:05:07: end import into table1 records (10082) I've posted about this before - one person will say that its my indexes getting rebuilt, others have said its disk io. I can never get a solid answer. If I disable the keys, do the import, then re-enable the keys, it takes just as long, if not longer. I have just about given up on finding a solution for this and just rotate my tables out regularly once the imports take over 5 minutes to process roughly 10,000 records -- George >>>-----Original Message----- >>>From: Jay Pipes [mailto:[EMAIL PROTECTED] >>>Sent: Wednesday, August 30, 2006 12:06 PM >>>To: Phantom >>>Cc: mysql@lists.mysql.com >>>Subject: Re: Degrading write performance using MySQL 5.0.24 >>> >>>On Wed, 2006-08-30 at 08:31 -0700, Phantom wrote: >>>> We have an application that stores versioned data in >>>MySQL. Everytime a >>>> piece of data is retrieved and written to, it is stored in >>>the database with >>>> a new version and all old versions are subsequently >>>deleted. We have a >>>> request rate of 2 million reads per hour and 1.25 million >>>per hour. What I >>>> am seeing is that as the DB grows the performance on the >>>writes degrades >>>> substantially. When I start with a fresh database writes >>>are at 70ms. But >>>> once the database reaches around 10GB the writes are at >>>200 ms. The DB can >>>> grow upto 35GB. I have tried almost performance related >>>tuning described in >>>> the MySQL documentation page. >>>> >>>> What do I need to look at to start addressing this problem >>>or this is how >>>> the performance is going to be ? >>> >>>Before getting into server parameters, is it possible to >>>take a look at >>>your schema and a sample of your SQL queries from the >>>application? That >>>would help immensely. 70ms for an UPDATE seems very slow... >>>and 200ms >>>is very slow. >>> >>>Cheers, >>>-- >>>Jay Pipes >>>Community Relations Manager, North America, MySQL, Inc. >>>[EMAIL PROTECTED] :: +1 614 406 1267 >>> >>> >>>-- >>>MySQL General Mailing List >>>For list archives: http://lists.mysql.com/mysql >>>To unsubscribe: >>>http://lists.mysql.com/[EMAIL PROTECTED] >>> >>> -- MySQL General Mailing List For list archives: http://lists.mysql.com/mysql To unsubscribe: http://lists.mysql.com/[EMAIL PROTECTED]
