> From: "Mike Spreitzer" <spreitze@parc.xerox.com>
> To: <wrec@cs.utk.EDU>, "Joe Touch" <touch@ISI.EDU>
> Cc: "Mike Spreitzer" <spreitze@parc.xerox.com>
> Subject: RE: Research Issues pre-draft
> Date: Tue, 23 Mar 1999 10:55:42 PST
>
> > Inter-replica communication
> > Demand Driven Acceleration Mirror Replication
> > [CODA] CODA
> > ?
>
> I hope I'm not stepping on any cross-town rivalries here, but if you mention
> CODA you should probably also mention Ficus. And I think Bayou deserves a
> mention too. The Bayou home page
As I discussed at the meeting, the objective of the draft
is to focus on the research _issue_, rather than the particulars
of any solution, where possible. The idea is to avoid exactly
this kind of 'contest'.
> And I'm not sure "Demand Driven Acceleration Mirror Replication" is the only
> issue these systems address.
Again, it's the issue that drives the I-D, not the systems used
as an example. The systems described need not be an exhaustive list
of examples either.
(this is not to justify mentioning CODA alone here; I point this
out only to avoid needing to list _every_ related project under every
issue. It suffices to include a representative, but not exhuastive, list)
> Indeed, it may miss a fairly large chunk,
> depending on how "mirror replication" is interpreted. If it means
> "master/slave replication", there should be another issue for "multi-master
> replication".
In this case, the issue might be "replication for increased
availability or incresed availability" (I don't think the two can be
otherwise separated - ??). The the master/slave vs. peer is a
technique to solve the issue, and CODA, Ficus, and Bayou are
representative solutions under each issue. However, since the
title of the ID is _Web_ Caching and Replication, it would be
useful to indicate how (or wheter) these systems have anything to
do with the web. The I-D isn't intended to be a book :-)
Joe
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:25 MST