Date: Thu, 30 Aug 2001 21:10:28 -0500 From: Mike Meyer <mwm@mired.org> To: "Kory Hamzeh" <kory@avatar.com> Cc: "Mike Meyer" <mwm@mired.org>, <questions@FreeBSD.ORG> Subject: RE: Performance tuning results Message-ID: <15246.61972.69707.967775@guru.mired.org> In-Reply-To: <003b01c13178$ad44fa60$14ce21c7@avatar.com> References: <15245.63001.898491.751110@guru.mired.org> <003b01c13178$ad44fa60$14ce21c7@avatar.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Kory Hamzeh <kory@avatar.com> types: > > > > -----Original Message----- > > From: owner-freebsd-questions@FreeBSD.ORG > > [mailto:owner-freebsd-questions@FreeBSD.ORG]On Behalf Of Mike Meyer > > > What's interesting is that IDE write caching helped more than the > > > softupdates. > > Why is that interesting? Softupdates caches things in the system > > memory to try and improve performance in a reliable manner. IDE disk > > caching caches things in the disks memory without worrying about > > reliability. One would expect the more reliable mechanism to be > > slower. > Because I assume that the system buffer cache is larger than the drive's > buffer cache. > > If you really want the extra speed - and don't care about reliability > > - you can mount your file systems async, softupdates off. If soft > > udpates are on, the async flag to mount is quietly ignored. That > > caches data in the system memory without regard to reliability just > > like the IDE disk cache does. > Why is IDE write caching less reliable than softupdates? They both basically > do the same thing: delay the write. The response is the same in both cases. Softupdates knows how the data in the cache is used in the file system, and orders the writes so that what's the drive actually claims is on the disk is in a consistent state. To the disk, all data is the same, so it will order the writes to get maximum performance. Note that all of the works softupdates does to make things reliable is wasted if the drive lies about what's actually written to disk - which is what the drive's write cache behaves. So if you're going to leave the drive write cache on, you should use async instead of softupdates. <mike -- Mike Meyer <mwm@mired.org> http://www.mired.org/home/mwm/ Independent WWW/Perforce/FreeBSD/Unix consultant, email for more information. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-questions" in the body of the message
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?15246.61972.69707.967775>