Re: MEMORY tables appalling insert performance
Posted by:
qihua wu
Date: December 31, 2009 01:43AM
You have 500 sql per second? that's 2ms for each sql. If the sql distribute evenly in the time slot, then there should be no lock as 2ms to insert/update is long enough for mysql to complete it. Your workload might be something like, idle for some time, and the burst of sql come in and each sql content for the table lock.
And How do you think innodb is not suitable? Innodb has cache, the dirty page doesn't need to be flushed to disk instantly after the page changed. Suppose the cache is large enough to cache all table/index (if one row is 1K, then 100K rows is only 100M), and have innodb flush the less the possible, much less PIO is required
Subject
Views
Written By
Posted
12500
October 26, 2009 11:46AM
6220
October 26, 2009 03:57PM
5581
December 31, 2009 01:30AM
5374
December 31, 2009 10:20AM
Re: MEMORY tables appalling insert performance
5850
December 31, 2009 01:43AM
5837
December 31, 2009 10:18AM
5349
December 31, 2009 08:54PM
Sorry, you can't reply to this topic. It has been closed.
Content reproduced on this site is the property of the respective copyright holders.
It is not reviewed in advance by Oracle and does not necessarily represent the opinion
of Oracle or any other party.