>>>>> On Thu, 22 May 2008 20:51:15 +0100 (BST), Alan Brown said:
> 
> On Thu, 22 May 2008, Martin Simmons wrote:
> 
> > Bweare that testing from /dev/zero might not give a realistic result if the
> > drive has compression switched on, because the phyical tape will hardly 
> > move.
> > It will test the interface card's speed nicely :-)
> >
> > It is probably better to make a large file containing real data, e.g. tar 
> > from
> > /usr/lib.  You could also compress the file with gzip, which will prevent 
> > the
> > drive from doing any compression.
> 
> Or use /dev/random.
> 
> It's a good idea to test limits by using /dev/zero or /dev/random and
> sources and /dev/null as sink to see what dd reports. This will help give
> some ideas of where your bottlenecks are.

You should use /dev/urandom on Linux, because /dev/random will continually
block until the kernel's entropy buffer is full.  Even /dev/urandom might be a
bit slow though.

__Martin

-------------------------------------------------------------------------
This SF.net email is sponsored by: Microsoft
Defy all challenges. Microsoft(R) Visual Studio 2008.
http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users