tag:blogger.com,1999:blog-14455177.post4971583817594841453..comments2024-03-07T18:57:25.977+01:00Comments on Mikael Ronstrom: MySQL 5.4 Patches: InnoDB Thread ConcurrencyMikael Ronstromhttp://www.blogger.com/profile/07134215866292829917noreply@blogger.comBlogger4125tag:blogger.com,1999:blog-14455177.post-49251311137030379472009-05-13T15:43:00.000+02:002009-05-13T15:43:00.000+02:00InnoDB thread concurrency usually works well when ...InnoDB thread concurrency usually works well when there is a significant scalability bottleneck which probably happens with inserts using autoincrement, haven't analysed this particular case.<br /><br />The scheduler I used was plain Linux using Red Hat.Mikael Ronstromhttps://www.blogger.com/profile/07134215866292829917noreply@blogger.comtag:blogger.com,1999:blog-14455177.post-62519164071263524092009-05-12T22:17:00.000+02:002009-05-12T22:17:00.000+02:00Nice article.
What type of scheduler did you rely...Nice article. <br />What type of scheduler did you rely on ? cfq or others ?Romain Ghttps://www.blogger.com/profile/10879424013679277912noreply@blogger.comtag:blogger.com,1999:blog-14455177.post-62783881867184522192009-05-12T20:54:00.000+02:002009-05-12T20:54:00.000+02:00Michael,
I've seen the opposite behavior (for MyS...Michael,<br /><br />I've seen the opposite behavior (for MySQL v5). I've had situations where innodb_thread_concurrency set to 0 caused bottlenecks on the master, especially on inserts where there is an auto_increment column, or a column that does not have a PK defined.<br /><br />Have you seen the same?<br /><br />ParthaPartha Duttahttps://www.blogger.com/profile/00972126296305304386noreply@blogger.comtag:blogger.com,1999:blog-14455177.post-27461248046137121792009-05-12T20:42:00.000+02:002009-05-12T20:42:00.000+02:00"It turned out I had found the optimal number at t...<I>"It turned out I had found the optimal number at the first attempt."</I>Which proves that you are a wizard?<br /><br />henrikhingohttps://www.blogger.com/profile/09201666166374161923noreply@blogger.com