lkml.org 
[lkml]   [1998]   [Dec]   [8]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
Date
From
SubjectRe: Dumb question: Which is "better" SCSI or IDE disks?
On Mon, 7 Dec 1998, C S Hendrix wrote:
> In message <Pine.LNX.4.04.9812070215060.3437-100000@localhost>, Christopher Smi
> th writes:
> > Well, based on this, your IDE drive is so fast it ever outperforms RAM!
> > ;-) (You are copying ~9 MB through a controller with 32 MB of cache.) My
> > suspicion is that your DPT RAID-1 controller with 32 MB cache is setup
> > with write caching disabled, as is the case for your SCSI hard disks. This
> > is one area where SCSI performance typically lags behind that of IDE,
> > because SCSI hard disk vendors typically setup their caches to not do
> > write caching by default (this can typically be changed, however). This is
> I noticed this was an option with the scsi tools under Linux, and
> I was wondering if it was prudent for me to turn on write cache on
> my drives. All of them support the option.
In general, if you are talking about a home or development system, I'd say
you should have cached writes enabled. In particular, ask yourself this
question: have you been worried enough to set chattr +S on some files on
your filesystem? If not, then you should DEFINITELY use write caching,
because the filesystem is already doing write caching. Even if you use
chattr +S, ask yourself if you are doing this to avoid application-level
issues (for example many mail programs require this) or if you are doing
it because you need to be able guaruntee that after a write has been
performed to a filesystem, that the write has actually made it to disk, so
that even if you have a major power failure, you know the data is safe
AT THAT SECOND. More than likely, the answer is no.

For a while, the notion circulated that having write caching disabled
avoided lost data from disk crashes, but this isn't really the case. Most
disk crashes occur from wear and tear on the drive, and as such occur
DURING the wear and tear (it's at least as likely the crash will happen
then as a second BEFORE the wear and tear). If anything write caching has
a chance to reduce problems with disk caches because it can (although
minutely) reduce disk wear and tear.

In general, I'd say, unless you know you need write caching disabled, you
probably should have it enabled.

--Chris


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 13:45    [W:0.117 / U:0.096 seconds]
©2003-2011 Jasper Spaans