Date: Thu, 06 Apr 95 23:37:49 EDT From: rsoles@terra.SIRIUS.COM (Roger L Soles) To: freebsd-hackers@FreeBSD.org Subject: Re: large filesystems/multiple disks [RAID] - DB Perf Message-ID: <9504070730.AA28143@SIRIUS.COM>
next in thread | raw e-mail | index | archive | help
If you're interested in achieving better database performance, you'd be better off to keep the spindles as seperate file systems and allow your database to optimize access to a segments which span are placed on each of the file systems. Depending on the database, there are almost always good ways to take advantage of the fact that you have multiple spindles... for instance, placing the rollback segment on a seperate spindle -- placing the audit log on a seperate spindle -- to name only a few. On Thu, 6 Apr 95 22:45:58 MDT you wrote: >> Yes. But mirroring gives additional throughput increase for reading >> like stripping (and decrease for writing unlike it :-( ). But from >> my experience big databases are much more often read than written, >> aren't they ? > >Yes, but I'd expect the agregate performance to stay about the same. > >Large databases don't allow predictive read-ahead because they >typically can't be modelled using a model that assumes locality >of reference. > >Unless your cache is significantly large relative to your database. > > > Terry Lambert > terry@cs.weber.edu >--- >Any opinions in this posting are my own and not those of my present >or previous employers. > > ----------> Roger L Soles PO Box 280785 San Francisco, CA 94128-0785
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?9504070730.AA28143>