Leasing Cluster

Robert G. Brown rgb at phy.duke.edu
Sat Dec 9 06:38:26 PST 2000

On Fri, 8 Dec 2000, Sam H wrote:

> I'd like people's opinions on the feasibility of setting up a cluster
with the sole purpose of leasing compute time to external companies.  Is
there a market?  Cost effective?

It's feasible, of course.  There is a market -- in fact such clusters
exist and can be cost effective to the users and profitable to the
owners.  However, the trick isn't in the building of the cluster, it is
in the usual:

   a) Marketing it -- doesn't do you any good to build it unless you
have a clear idea where you want to sell its services and are sure that
you don't face entrenched competition in that marketplace.  Don't
underestimate the difficulty or importance of sales.  The streets are
currently littered with the bodies of .com's that thought the "if we
build it, they will come" business model held water.

   b) In order to market, you'd better come up with a very specific
application niche that you want to push.  You're a lot better off
starting out with focus in just one reasonably vertical market than
trying to get a little business here and there.  Don't count on
advertising "generic compute cycles for sale" and expect a customer to
come to you with an application.  Remember, compute cycles are dirt
cheap; that's not really what you are selling.  You're selling savings
on overall infrastructure, the ability to enter a fixed cost in a budget
estimate, you're selling consultative expertise.

In once sense this is what we do now.  However, we've been very hesitant
about branching out to make this a business focus.


> Sam Harper
> Tibernia
> mudguy at speedfactory.net
> 770-401-5518

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 mailing list