MySQL Forums
Forum List  »  Performance

Advanced Search
New Topic

Re: sync_binlog=1 and innodb_flush_log_at_trx_commit = 1 killing write performance
Posted by: Rick James
Date: June 26, 2014 06:19PM

> The question we have is, considering our replication strategy, would it be safe to run with sync-binlog = 0
(or a high number like 100)

"Safe" is relative. Flushing the binlog will make recovery from a crash a little simpler. It will rarely be any safer.

> AND innodb_flush_log_at_trx_commit = 0?

Consider using "2", not "0". It will be much faster, but will put you at risk of loosing one second's worth of writes.

Better yet, get a hardware RAID controller with a Battery Backed Write Cache -- then your writes can be "instantaneous" and =1 becomes a good idea.

> under the impression that 5.6 and newer mysql drastically improved these sorts of issues.

I don't now of any.

Navigate: Previous Message• Next Message
Options: Reply• Quote


Subject
Views
Written By
Posted
sync_binlog=1 and innodb_flush_log_at_trx_commit = 1 killing write performance
6314
Bill Plimpton
June 23, 2014 04:52PM
Re: sync_binlog=1 and innodb_flush_log_at_trx_commit = 1 killing write performance
1613
Rick James
June 26, 2014 06:19PM


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.