*BSD News Article 11445


Return to BSD News archive

Received: by minnie.vk1xwt.ampr.org with NNTP
	id AA1906 ; Tue, 23 Feb 93 15:03:23 EST
Newsgroups: comp.unix.bsd
Path: sserve!manuel.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!sun-barr!cs.utexas.edu!geraldo.cc.utexas.edu!portal.austin.ibm.com!awdprime.austin.ibm.com!guyd
From: guyd@austin.ibm.com (Guy Dawson)
Subject: Re: [386BSD] What SCSI controllers _are_ supported?
Originator: guyd@pal500.austin.ibm.com
Sender: news@austin.ibm.com (News id)
Message-ID: <C2uzsA.2542@austin.ibm.com>
Date: Mon, 22 Feb 1993 16:43:22 GMT
References: <1993Feb17.214948.9390@fcom.cc.utah.edu> <C2no1H.1JDo@austin.ibm.com> <1993Feb19.033337.13588@fcom.cc.utah.edu> <C2q1Lu.7LJ@sugar.neosoft.com>
Organization: IBM Austin
Keywords: 386BSD SCSI
Lines: 32


In article <C2q1Lu.7LJ@sugar.neosoft.com>, karl@NeoSoft.com (Karl Lehenbauer) writes:
> In article <1993Feb19.033337.13588@fcom.cc.utah.edu> terry@cs.weber.edu (A Wizard of Earth C) writes:
> >My opinion unsubstantiated by testing:  A cached controller is useless
> >for anything other than slowing initial transfer rates unless the controller
> >cache is *much* larger than the UNIX cache *and* the transfer rate between
> >the controller cache and memory is the same as that between memory and
> >memory.  Spend your money on something you can see, like a big monitor.
> 
> I agree with Terry (but also did not test).  Also caching controllers
> change the order that writes are performed.  It is important, at least
> for System V file systems, that blocks go out in the order the system wants
> them to go out in.  The system thus ensures that the data of a file is
> out there before the inode is updated, and the inode is out there before
> the directory is updated, and so on, to insure that things aren't left
> weird and incomplete if the system crashes.  Thus caching controllers
> that know nothing of the filesystem reduce the robustness of the filesystem.
> An UPS is mandatory, IMHO, if you are using a caching controller.
> -- 


The above is a *very* good example of the type of advantage that a Unix disk
cache has over a hardware cache. Unix caches the file system, a disk cache
caches the disk. The difference being that the Unix cache 'knows' about
the data it is caching, what it is, what it is used for, how it is used etc.
The hardware cache just knows about some disk blocks.

--------------------------------------------------------------------------------
Guy Dawson - Hoskyns Group Plc.
        guyd@hoskyns.co.uk  Tel Hoskyns UK     -  71 251 2128
        guyd@austin.ibm.com Tel IBM Austin USA - 512 838 3377
"Knolege is powef, Speling is unimportnt" via Pete W. De Bonte