I've Looked at Clouds from Both Sides Now
Toby Lehman
Almaden Services Research
IBM Almaden Research Center
650 Harry Road
San Jose, CA 95120
toby@almaden.ibm.com

In the year 2002, at the IBM Almaden Research Center, Jamie Kaufman and I 
started an autonomic grid computing project called OptimalGrid whose goal 
was to create a compute grid infrastructure that was flexible, 
self-healing and able to support general-purpose scientific computing. 
After an awkward start with scientific applications (it turns out there’s 
almost no money in scientific applications), we had some minor successes 
with configuring our grid to be a flexible, dynamic compute infrastructure 
for massively multiplayer online gaming.  We showed how our system could 
detect the computing needs of the distributed game application and 
dynamically allocate, then deploy, additional servers when there was a 
need for increased capacity, and reclaim servers back into the shared 
server pool when there was a decreased capacity.

Since the dynamic backend capability was somewhat new, we attracted the 
attention of various gaming and hosting companies.  We gave many 
presentations, with basically the same result.  “Ah, excellent.  Very 
Interesting!”  we heard over and over.  Next, “we’ll get back to you when 
we’re ready to move on this.”   Then, silence.  Eventually, we gave our 
last presentation.  Then our project ended.

There are many reasons why the interest in our grid system waxed and 
waned.  Clearly, there could have been faults in our design or technology 
which caused visitors to run for cover.  However, the lack of any other 
significant deployment of similar technology points to a more obvious and 
fairly obvious answer.  Business is fickle.  Furthermore, business that 
relies on new technology is both fickle and paranoid.  Regardless of the 
“cool factor” of the technology demo, in order for a company to 
significantly adopt a new technology, they just possess courage, vision 
and a strong commitment to see it through.  And, there must be no worries, 
no skeletons in the closet.

That brings me to Cloud Computing.  Having kinship with grid, I see many 
of the same thrills and chills.  How Wonderful!!  It will solve all of our 
problems!  It’s FREE ! 

There are lots of hand-waving and interesting demos, but in the end, I 
suspect that Cloud Computing will be expected to answer the same questions 
we never quite answered in the grid project.  There are many questions, 
but here's my favorite:  What about security?  How exactly does the user 
trust the cloud, and how does the cloud trust the user?  Would a company 
want to put its important financial data in the hands of a cloud machine? 
Grid Computing at least had the scientific or game demo, which could 
illustrate the wonderful speedup of multiple computing machines, without 
the hassle of addressing real world problems such as security.  Cloud 
computing, however, is promoting its general purpose nature – everything 
can be part of the cloud.  As a result the problems get harder, the 
questions get more pointed, and so far, the answers, or lack thereof, seem 
to remain the same.

Toby Lehman

Office:       (408) 927-1781    (IBM tieline 8-457-1781)
ARC Fax: (408) 927-2100    (IBM tieline 8-457-2100)
Home Page:   http://www.almaden.ibm.com/cs/people/toby