[Beowulf] [OT] HPC and University IT - forum/mailing list?
Robert G. Brown
rgb at phy.duke.edu
Tue Aug 15 14:12:59 PDT 2006
On Tue, 15 Aug 2006, Bill Rankin wrote:
>> Are there problems so specific to the higher education realm
>> that you think they'd benefit from their own forum?
> Not so much that would benefit from their own forum, but there are a lot of
> issues that are not directly related to the construction and technical
> operation of clusters, but rather more administrative/organizational in
> Things like:
> - integration of a cluster into a larger University IT infrastructure
> (storage, authentication, policies, et. al.)
> - funding models (central funds, grant based) both for equipment and
> - centralized research IT versus local departmental/school support.
> - education and training
> - deployment issues (who pays F&M?)
> - sustainability and growth
> Some of these topics have content that would be appropriate for this list,
> some not-so-much. There are also some people I know that would be interested
> in participating in the conversation, but not about general Beowulfery.
This is a key question (and answer). There are issues of critical mass
for a successful list to consider. There are also alternative formats
-- e.g. wiki instead of actual mailing list. Finally there are issues
of sponsorship and linkage -- it is easier to get to critical mass if
there is a mechanism other than google for feeding people and certain
threads of discussion over to the new venue. Possibilities include
crossconnecting/linking the new resource with an existing resource to
get some synergy, hijacking threads on the beowulf list that are
sufficiently "academic" that they are as noted slightly OT for what most
of the list membership cares to here (if there is any way to note that),
periodic advertisement on the beowulf list (in the form of gentle
invitation to move threads vs hijacking:-).
>> So if it were a democratic process, I'd vote to keep it here, but if a
>> new list/site is created, I'll go wherever the information is.
> I just wanted to try and avoid throwing a lot of off-topic posts onto this
The list in the past has been pretty tolerant about the width of topic
represented by its many many threads. It is pretty easy to ignore any
given thread after all, and overall the signal to noise level on this
list has remained fantastically high for roughly a decade. It's also
hard to second-guess what list members ARE interested in. The
commercial hardware or software folks sell to -- universities, among
other places. Many many cluster owners and users and administrators on
list work at/in universities. Even total corporate cluster
builders/admins/users have SOME overlap with universities although their
needs and funding are likely to be different in lots of places as well.
I'd wait for somebody to complain, personally, before concluding that
the beowulf list can't handle the traffic you describe.
The bigger problem may be in the other direction. Are there individuals
and groups that you think would participate in the new proposed forum
who would NOT join to beowulf list to do so, for whom the bulk of
beowulf traffic itself is "noise"? If you are thinking of a "list for
university deans" or members of research support offices or departmental
administrators -- places where they can discuss funding models that
would have no interest whatsoever to cluster builders (and vice versa,
where they don't care about coarse vs fine grained, only how to set up a
cost-sharing mechanism that granting agencies will support and have it
administerable and accountable should they get audited) -- then yeah, I
think a new list or other venue would be very useful.
There you WILL have to deal with the issue of keeping the interest of a
stable of real cluster experts, because of course the list will not be
useful unless its ultimate conclusions DO include all sorts of things
derived from a detailed knowledge of interconnects, power generation by
nodes, coarse vs fine grained requirements... so SOMEBODY will have to
be able to do deanspeak as well as clustertalk and enjoy it enough to do
it voluntarily. Or get paid. Or both. Then you still have the problem
of building it and having them come, but you'd have a unique and
important product and with enough feeders and some solid content online
one way or another and maybe an associated consulting service you might
make it go...
>> many subscribers are on this list now anyways? The www.beowulf.org
>> website says, "Over 3,000 researchers, engineers, students, and software
>> developers have participated in these online discussions since 1996."
>> but it doesn't say how many are subscribed now.)
> That would be an interesting number to know.
It would indeed. I'll say from some offline experience that there are
many, many more "members" than actually participate in discussion
threads. I end up in offline conversations or real human conversations
with folks who have lurked on the list for years, or used its valuable
content via google from the archives, and are contacting me just on some
specific point. Then there are digested remailings of selected list
threads and more. There is a rather large virtual audience outside of
the participatory one...
> Thanks for the feedback,
> Bill Rankin, Ph.D.
> Director, Cluster and Grid Technology Group
> Center for Computational Science, Engineering and Medicine
> Duke University
> wrankin at duke.edu
> ph: (919) 660-6561
> fx: (919) 660-7070
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit
Robert G. Brown http://www.phy.duke.edu/~rgb/
Duke University Dept. of Physics, Box 90305
Durham, N.C. 27708-0305
Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb at phy.duke.edu
More information about the Beowulf