Subject: Grid computing, and dept. notes
From: Keunwoo Lee (klee@cs.washington.edu)
Date: Mon Oct 22 2001 - 16:17:27 PDT
Apparently this "grid computing" buzzword has a lot more muscle behind it
than I thought. The European High Performance Computing & Networking
conference (HPCN) 2001 this summer made "grids" one of the major themes
of their workshop:
http://www.science.uva.nl/events/HPCN2001/hpcn/programme.php
(Unfortunately, no papers linked from this page.)
I found this stuff while doing some background searches on Java Grande and
the use of Java for high-performance computing. I had heard the phrase
"grid computing" exactly zero times until this past summer, so I was
pretty surprised.
What is "grid computing"? As far as I can tell, it's a generalization of
Seti@Home etc., plus a lot of fancy verbal hoo-ha to impress people. The
computational power in distributed systems makes natural scientists drool,
and they need computer scientists to help them get at it.
I believe the NSF is funding a boatload of this stuff; one group's site:
http://www.globus.org/research/papers.html
Our department has strong people in systems, programming systems, and
parallel systems. We are a natural place to be doing this kind of stuff,
if anybody were interested in it, and if we could get the relevant people
to cooperate. Perhaps sometime this year we should approach professors
and new students in the systems and ZPL groups and see if any of them are
interested. Not something we can do a grant proposal for this year,
obviously, but in the longer term, contingent on interest, of course.
To make this relevant to 590L: Cardelli's mobile computation paper has a
"webbish" applications flavor, so it may lead you to think about applets,
mobile agents, etc. shuttling around the web. High-performance computing
is another application to keep in mind, with different demands and
tradeoffs.
Oh, last thing: I did not know this until today, but Steve (Gribble) and
Andrew (Whitaker) did some lightweight distributed virtual machines
brainstorming this summer, and had a poster at SOSP about it:
http://denali.cs.washington.edu
It has an OS flavor (basically, their "VM" looks to me like an exokernel
with a twist). Nevertheless, I wish I had known about this. I think
perhaps our group (well, maybe just me) needs to communicate more with
other groups in the department. If we're going to learn about programming
support for evolving/distributed systems, we had better talk to people who
build distributed systems.
~k
_______________________________________________
Cecil mailing list
Cecil@cs.washington.edu
http://majordomo.cs.washington.edu/mailman/listinfo/cecil
This archive was generated by hypermail 2b25 : Mon Oct 22 2001 - 16:18:03 PDT