Return to BSD News archive
Xref: sserve comp.unix.bsd:8043 comp.benchmarks:2447 comp.arch:28420 comp.arch.storage:711 Path: sserve!manuel.anu.edu.au!munnari.oz.au!network.ucsd.edu!usc!sol.ctr.columbia.edu!ira.uka.de!math.fu-berlin.de!mailgzrz.TU-Berlin.DE!gmdtub!bigfoot!tmh From: tmh@keks.first.gmd.de (Thomas Hoberg) Newsgroups: comp.unix.bsd,comp.benchmarks,comp.arch,comp.arch.storage Subject: Re: Disk performance issues, was IDE vs SCSI-2 using iozone Message-ID: <TMH.92Nov21102943@keks.first.gmd.de> Date: 21 Nov 92 09:29:43 GMT References: <1992Nov7.102940.12338@igor.tamri.com> <36794@cbmvax.commodore.com> <1992Nov10.170022.21624@igor.tamri.com> <36994@cbmvax.commodore.com> Sender: news@bigfoot.first.gmd.de Followup-To: comp.unix.bsd Organization: GMD-FIRST, Berlin Lines: 33 In-reply-to: jesup@cbmvax.commodore.com's message of 12 Nov 92 05:24:34 GMT In article <36994@cbmvax.commodore.com> jesup@cbmvax.commodore.com (Randell Jesup) writes: >> Write-buffering (starting to be commonly available as an option on >>SCSI drives) can help in avoiding slipping revs. >While some speed can be gained by this practice, all ability to handle >error conditions responsibly is generally lost. Yes, write-buffering does lose some error recovery chances, especially if there's no higher-level knowledge of possible write-buffering so filesystems can insert lock-points to retain consistency. However, it can be a vast speed improvement. It all depends on your (the user's) needs. Some can easily live with it, some can't, some need raid arrays and UPS's. Hmm, I've always thought of raid arrays as a means of enhancing performance and protection against data loss due to drive failures. UPS's become a necessity if write buffering controllers (or disks) break file system hardening strategies by reordering writes or asyncing synchronous writes. Two strategies come to mind to ease the write performance bottle neck: Motors and generators are rather similar and the kinetical force of a couple of disks spinning at 3600 revs should be enough to let the drive electronics finish writing some 20 blocks or so (the engineering cost might be a bit high, though). Using CMOS memory and a backup battery is another. With some 512K of battery protected CMOS RAM you should be able to to increase write performance significantly without foregoing file system consistency. --- Thomas M. Hoberg | Internet: tmh@first.gmd.de 1000 Berlin 41 | tmh@cs.tu-berlin.de Wielandstr. 4 | Germany | BITNET: tmh@tub.bitnet +49-30-851-50-21 |