Date: Fri, 29 Apr 2005 16:00:57 +0100 From: Brian Candler <B.Candler@pobox.com> To: killing@multiplay.co.uk Cc: freebsd-current@freebsd.org Subject: Re: Very low disk performance Highpoint 1820a Message-ID: <20050429150057.GA93707@uk.tiscali.com> In-Reply-To: <42722CB9.7050702@corp.grupos.com.br> References: <069901c54bfd$2967ba40$7f06000a@int.mediasurface.com> <42722CB9.7050702@corp.grupos.com.br>
next in thread | previous in thread | raw e-mail | index | archive | help
> >5.4-STABLE Highpoint 1820a RAID 5 ( 5 disk ) > >dd if=/dev/da0 of=/dev/null bs=64k count=10000 > >10000+0 records in > >10000+0 records out > >655360000 bytes transferred in 13.348032 secs (49097875 bytes/sec) > > > >5.3-RELEASE Highpoint 454 RAID 5 ( 4 disk ) > >dd if=/dev/da0 of=/dev/null bs=64k count=10000 > >10000+0 records in > >10000+0 records out > >655360000 bytes transferred in 20.410034 secs (32109697 bytes/sec) > > > >5.2.1-RELEASE Intel ICH3 UDMA100 ( 1 disk ) > >dd if=/dev/ad0 of=/dev/null bs=64k count=10000 > >10000+0 records in > >10000+0 records out > >655360000 bytes transferred in 11.142405 secs (58816745 bytes/sec) > > > >Obviously something is seriously a miss here somewhere as > >both the RAID 5 arrays a producing less throughput than > >a single disk. > > > >Where do I start looking? Just some ideas: - try reconfiguring your array as 5 separate disks, and dd from a single disk. If it's still slower than your ATA disk, then it's probably SCSI transfers or controller I/O bandwidth which are slowing things down. If you get equal or better performance than your ATA disk, then it becomes likely that the RAID configuration is the bottleneck. - try reconfiguring your array as two mirrored pairs, and do the same test again. RAID5 isn't necessarily a good choice for high-performance applications; a single block write requires two reads and two writes (to the target data disk and the parity disk), and therefore write access to the array is likely to be *slower* than to a single disk. You might be better off with mirroring. It handles random writes better, and you may get double the random read performance since there are two copies of all the data. RAID5 is acceptable if your application is mostly read-only though (which your dd test is, of course) Brian.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20050429150057.GA93707>