==> On Fri, 26 Aug 2005 21:33:58 -0400, David E Ehresman <deehre01 AT 
LOUISVILLE DOT EDU> said:


> Uhh, how about for restore performance when it matters most, during a 
> disaster!


I come down on the other side of this, myself.  Biggest reason is, you just
can't afford to collocate the offsites.  Imagine sending one tape per node
offsite, every day.  If you've got some sort of long-haul fiber solution, and
can expose the physical volume granularity at the remote site to your local
server, then it would be very useful to collocate.

So in your site-wide disaster, you are going to have a mishmash anyway.


On the other hand, the media-failure case is much more common.  In those
cases, I find the noncollocated copy pool to be a reasonable price to pay for
the savings in slots and tapes.


All these calculations might be different with collocation groups; hopefully
in a week or two I'll be considering that as a concrete problem. (chortle)


- Allen S. Rout