moving on.

David Jeske
Thu, 17 Dec 1998 23:51:03 -0800

On Thu, Dec 17, 1998 at 08:23:00PM -0500, Maneesh Yadav wrote:
> The one thing I've always wonderd about distributed storage is how
> is where is the information regarding where the data physically is,
> held?  I mean there are a lot of nice paragmatic things about such a
> system, but in the end not only do you sometimes want to know where
> a piece of data is, you want to put data in a specific place.

Certainly. You often want to, _for a given target environment_ specify
where a piece of data should go. I think the Aspect oriented people
are going the right direction by trying to separate the above
'optimization information' from the 'functional algorithm'. They are
not the same thing. If you want to run a functional algorithm 2 years
later, it's likely that your original optimization assumptions will
not hold for the hardware you want to run it on, and maybe everything
will run better if they are just discarded.

David Jeske (N9LCA) + +