[OpenAFS-devel] Re: Backups using commercial products

[Mitch Collinsworth]

On Wed, 3 Jan 2001, Ken Hornstein wrote:

> >> What I wonder is - how hard would it be to access the contents of an AFS
> >> volume from a file server, without going through AFS itself - i.e. by direct
> >> file access - sortof a really fast 'vos dump' equivalent.
> >
> >Good question.  Now that we have the source I'm sure it would be
> >possible to figure out the structure and read it directly from
> >disk.  Maybe somebody's even done this already.
> 
> I've done part of that already (dumptool takes a reasonable stab at
> documenting the dump structure).  However ... I'm not completely convinced
> that it will be _that_ much faster, since I suspect the bottleneck
> is reading the data off of the disk (lots of seeks).  But hey, I could
> be wrong.

We seem to keep circling back to this same issue.  Volume dumps are
inherently slow because of all the disk seeks required.  If this
is really the problem then maybe something that needs to go on the
OpenAFS project list is looking at ways of improving the on-disk
volume format.

-Mitch