 |
|
DAILY NEWS AND INFORMATION
FOR THE GLOBAL GRID COMMUNITY / OCTOBER 13, 2003: VOL. 2 NO. 41
|
Special Features:
CASSATT ASSEMBLES DREAM TEAM FOR
INTERSECTION OF WEB, GRID
By Rachel Chalmers For the451.com
Bill Coleman has pulled the wraps off his latest venture, Cassatt. It is a
dream team put together to exploit the opportunities created by Web services
and Grid computing. Cassatt plans to offer autonomic computing software based
on SNMP, Linux and J2EE, and to extend its features to Windows via standard
interfaces. Direct support for Windows is coming soon. There also may be ports
to proprietary Unix systems, depending on customer demand. The company has
chalked up an undisclosed amount of seed funding from global private equity
firm Warburg Pincus, which did very well as a result of Coleman's last
venture. He is the co-founder and former chairman and CEO of BEA Systems.
Impact Assessment
The Message
Take Bill Coleman, the B in BEA, add the creator of programmer's essential
versioning tool CVS, and throw in veterans of Sun, IBM, Oracle and SGI. You've
built a company to watch.
Competitive Landscape
Cassatt plans to complement, and not compete with, software from market
makers
like BEA, BMC, HP, Oracle and Veritas. But IBM, Mercury, Quest, Platform and
Sun are also keen to exploit SOAP and the Grid, as are several hungry
startups: Centrata, Ejasent, MetiLinx and The Mind Electric, to name a
few.
The451 Assessment
Details on what Cassatt intends to do are deliberately sketchy, but the
quality of the executive team is arresting.
Context
The company employs 15 people, but what a team they are. To miss Coleman's
achievements in the Web application server market you would need to have been
asleep for the last seven or eight years. His CTO, Brian Berliner, boasts
equally remarkable credentials.
Besides having written CVS, which despite its flaws is the default source
code
versioning system for software projects, Berliner headed the engineering team
at Sun that wrote Solaris. Yet another big name, Steve Oberlin, has already
come to Cassatt via acquisition.
The company is based in Menlo Park, Calif., with offices in St. Paul,
Minn.,
and Colorado Springs, Colo. It's named for the American Impressionist painter
Mary Cassatt, who followed Georges Seurat in the pointillist tradition. The
rationale is that Mary Cassatt's art, like Cassatt's software, joins the
dots.
Technology
Cassatt's working theory is that Moore's Law has not kept pace with the
complexity of software. Network operators don't need more equipment; they need
more control. Cassatt proposes to give it to them by exploiting the
convergence of Grid and Web services standards. In particular, explains Vice
President of Product Marketing Dave McAllister, Cassatt hopes to exploit late
binding across the network. As he sees it, the IBM era saw a single
application running on a single mainframe –- complex enough, but still
manageable. The appearance of n-tier applications talking to each other over
CORBA and DCOM greatly increased the difficulty of managing business software.
Now with J2EE mature and Web services coming into their own, almost all
applications need to request information from elsewhere on the network.
Cassatt has identified four levels of abstraction that need to be
addressed.
First is the massive set of interconnections created by applications that
routinely talk to each other. Second is the presence of both distributed and
detached storage, making file system expertise a must. Third is a move from
large-scale to small-node clusters, and the corresponding increase in the
number of CPUs. Fourth is the fact that applications themselves are now being
modified to take advantage of the distributed world. McAllister argues that
the incumbents -– among whom he inlcudes BMC, HP, Oracle and Veritas -– offer
software to solve problems at one or two of these levels. Cassatt aims to
complement what they can do by tackling all four levels at once. It sounds
frustratingly vague, but McAllister assures the451 that the software exists
and that it is already running at several (unnamed) sites.
Strategy
Cassatt remains very close to Coleman's alma mater, BEA Systems: the
company
plans to resell BEA software as part of its strategy for J2EE and Web
services. Given BEA's phenomenally successful track record with acquisitions
–- WebLogic itself came to the company via a merger –- it should come as no
surprise that Cassatt has already bought its first company.
Unlimited Scale Inc (USI) was a Linux clustering company formed by a group
of
ex-Cray Research employees including Steve Oberlin, chief architect of the
Cray T3E. Funded by HP, Sandia National Laboratory and Quatris Fund, USI
specialized in high-performance and technical computing (HPTC). The team will
operate as an independent technology division within Cassatt, and together
with the company's Colorado Springs division will form the basis of a product
development organization. HPTC won't be Cassatt's primary focus, but
executives say the company will continue to support USI's customers.
Competition
Companies taking on the convergence of SOAP and the Grid are as thick on
the
ground as autumn leaves. IBM is spearheading the Open Grid Services
Architecture, a standards effort aimed at uniting the Globus Toolkit with the
SOAP stack. Oracle sees the Grid as an extension of its Real Application
Cluster. BEA, BMC, HP, Mercury, Quest, Platform and Sun are buying or building
tools to manage distributed environments as if they were single systems, and
to express the results as Web services. Among startups, Centrata, Ejasent,
MetiLinx and The Mind Electric have bet their businesses on Web services and
utility computing. Cassatt CEO Coleman likes to say that competition makes
markets, but he has his work cut out for him here.
Courtesy www.the451.com
|