I didn't mean to seem so restrictive -- but I didn't have a better
umbrella term to encompass mirrors, replicators, caches, proxies,
and <insert your favorite term here>.
Here's my conception, and you can tell me where I'm baked:
+---------+ (1) +----------------+ (2) +---------+
| clients | <-> | intermediaries | <-> | servers |
+---------+ +----------------+ +---------+
(Hmm... intermediaries seems a little stilted, but it works for the
time being). The main problem right now is that (1) is manually
configured and should be autoconfigured.
Of course, there are different types of intermediaries, so that does
imply some architecture and requirements...
> > The goal of this working group is to define and specify solutions in the
> > following areas:
> >
> > - Proxy architecture and requirements,
> >
> > - Proxy autodiscovery protocols and mechansims,
>
> why should this be limited to proxies?
> proxies are of course needed for backward compatibility,
> but direct client access to the replication oracle is highly desirable.
>
> Keith
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:25 MST