[Beowulf] [OT] HPC and University IT - forum/mailing list?
jmdavis1 at vcu.edu
Tue Aug 15 17:44:26 PDT 2006
Mark Hahn wrote:
> huh? what value does big-A have to add here? the correct queueing
> system is
> the one that is cheap, low-maintenance, efficient, easy to use, etc.
> those are things that users and sysadmins know, not behind-desk-sitters...
Difference of definition here. I believe that Big-A administration is
how to best manage the resources of Technology to meet everyones needs.
In that sense, a submission system is very important because it lets us
leverage those resources.
I can understand frustration with the Big-A. Bean counters can be
frustrating. A priority of the day mentality can as well.
My background is film. Filmmaking is all about solving problems.
Twenty-one years ago, my film department got some computers to use for
animation and whatever else they could be used for. At the time, I was
paid to run the department's editing lab. So I learned about the Amiga
and how to use it to do more than anyone thought it would. That led to
writing some code to make a Sun box run an edit list and tell a
controller that had a serial interface what to do. That was 20 years
ago. For the past nine years, I've worked mostly on HPC and research
computing. We've worked our way from vax to irix to solaris and
clusters. We've gone from 4 processors to 500 dedicated to research.
These days at least half of my time is Big-A administration, managing
people and working with departments to get work done. This focus has led
to growth and personal trust from these departments. They know that we
won't always be able to give them what they want. But they believe that
they will get as many of their needs met as we can.
For example, I can set up one cluster, with multiple queues and
priorities that will serve multiple owners with a single head node,
address, networking scheme, storage scheme, and connect it to the
University's centralized backup system. This means more nodes for work
and less for administrivia. It means that we don't have multiple batch
systems on multiple headnodes running the jobs for many individual
machines. Is this solution always best for any one individual? No. Does
it let the departments get more research done? Yes.
So, even now, my job is about solving problems. But the end result is
more important than any film. The end result may be something that saves
a life, or improves ones standard of life. The end result may even
change the world.
Mike Davis (feeling far too wordy and poetic)
More information about the Beowulf