[Beowulf] What's the category of Beowulf among Clusters?

[Robert G. Brown]

On Fri, 19 Dec 2008, Mark Hahn wrote:

>> what's the category of Beowulf like clusters?
>
> beowulf is compute clustering using mostly commodity hardware and mostly 
> open-source software.

And if you want to be really picky, it should be an architecture that
"looks like a supercomputer" in the sense that it has an inside network
containing nodes and a dual-network head node that "is the cluster" as
far as the outside networks are concerned.  That is, the original
beowulf was designed to run relatively tightly coupled and synchronous
code.

However, at this point beowulf is like kleenex -- a brand name that has
become synonymous with more than its strict original definition.  People
will refer to nearly any HPC cluster as "a beowulf", and quite a few
people misuse the term to include some classes of HA clusters as well.
It may be that in the end beowulf simply means "compute cluster", but on
this list we try to rule out HA most of the time because it has some
different concerns (as well as some concerns that are the same).  My own
clusters, for example, have always been much more in the distributed
non-beowulf flavor, but they are definitely HPC and so I get to play.
Grids are even further away from strict beowulfs, but we've certainly
had many grid discussions on list.  So this is really the "HPC compute
cluster list" with rare nods to an HA topic, with the understanding that
the most "interesting" list discussions focus on parallel computation
coding (tools, languages etc), cluster design, and cluster networking
advanced or mundane.

    rgb

> _______________________________________________
> Beowulf mailing list, Beowulf at beowulf.org
> To change your subscription (digest mode or unsubscribe) visit 
> http://www.beowulf.org/mailman/listinfo/beowulf
>

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