DIS: slowly changing objects

[Francois-Rene Rideau]

Mike asked me what a slowly changing object was.
Well, there was a discussion about distributed systems
in the old MOOSE mailing list, and one of us pointed to
some articles in the net (I'll try to retrieve it).
Among these, techniques for efficient distributed systems
were described.
One thing that they measured was that most objects didn't change much,
or didn't change at all (which would be much easier for us if we used
some functional programming model with as few border/side effects as
possible, like in the STAPLE system).
So they found that by copying data, and have multiple copies around the
distributed system, you could accelerate accesses, save net bandwith,
and increase system independence toward link-quality/reliability altogether.
They also demonstrated efficient methods to buffer non-constant data in
a consistent way.

When I have time, I'll parse the 1.5MB old MOOSE archive to locate
these articles.