Hi Alan, Strictly on a disk I/O issue, I would test diff raid/disk configs and use Iozone with the wks output flag which generates some nice Exel spread sheets and graphs. I was at a job years ago were I used a combo of strace, ptrace and some other mysql debug utils to determine why the DB/server was unusable at certain times and why. I used to employ the 'throw pasta against the wall to see which sticks" approach but found it stressful and use Iozone for disk throughput analysis. On Oct 9, 2009, at 10:55 AM, Alan McKay wrote: >> Since you stated that your system is only having trouble at writes, >> doing batch inserts, I would look at how you are doing these batch >> inserts and see if you can optimize this by using techniques to delay >> key writes or adjust locking mechanisms on the insert. Some databases >> have special insert from file or bulk insert operations that are much >> faster than running many single row inserts - be sure you are taking >> advantage of these methods. > > Yeah, our DB geeks have already said a few weeks ago that they think > we are doing these bulk updates in a pretty inefficient manner, so > that's something else we plan to optimize. > > > > -- > “Don't eat anything you've ever seen advertised on TV” > - Michael Pollan, author of "In Defense of Food" > _______________________________________________ > CentOS mailing list > CentOS at centos.org > http://lists.centos.org/mailman/listinfo/centos