 |
|
DAILY NEWS AND INFORMATION
FOR THE GLOBAL GRID COMMUNITY / SEPTEMBER 22, 2003: VOL. 2 NO. 38
|
Special Features:
GGF CHAIR RESPONDS TO
ORACLE
Following is a response by the chair of the Global Grid Forum to the recent
announcement by Oracle that it intends to set up its own consortium to define
Grid-computing standards, a move that could create tension with the Global
Grid Forum, of which Oracle is already a member.
THOUGHTS ON COMMERCIAL GRID STANDARDS
by Charlie Catlett, Chair, Global Grid Forum
This week, Oracle announced the concept of a new Grid standards
organization
aimed at commercial standards, cooperating rather than competing with extant
standards organizations, including Global Grid Forum. I've not yet had a
chance to speak with anyone at Oracle, nor have any of the other GGF leaders,
so without more information it's not yet clear what will be the focus of such
a consortium.
Is another Grid-related standards group needed? It depends on what the
focus
would be, and how that focus relates to ongoing work. We're following up with
our friends at Oracle to see what they are thinking. My hope is that a dialog
would show Oracle that building strong, commercially useful software and
services based on open standards is something that is at the heart of the work
being done by both research and commercial leaders in GGF. If, after a dialog,
it turns out another standards body is necessary, then the community will have
had the benefit of coordinating from the very beginning!
The coverage of the Oracle announcement, and comments by pundits, has made
a
number of statements about the Global Grid Forum which are worth closer
examination. Unfortunately, there has been a combination of inaccurate
information and suggestions that there may be a conflict between the research
and commercial communities who we see collaborating closely in GGF. The
primary arguments in favor of a new, commercial Grid standards effort have had
to do with the difference between science/research and commercial approaches
to computing, coupled with a characterization of GGF as primarily
non-commercial.
Is GGF primarily non-commercial? 40 percent of GGF participants are from
industry, and two thirds of GGF sponsoring organizations are commercial
companies. There are 46 companies who are sponsor members of GGF, and at
every-level of leadership within GGF there are commercial participants from
many of these companies. The top contributors to our sponsor membership
program this year include Cisco Systems, Hewlett-Packard, IBM, Microsoft,
Qwest Communications , Silicon Graphics and Sun Microsystems. These and other
software industry leaders, including Oracle, are participating in GGF, but we
have a host of companies beyond the 46 sponsors that are smaller software and
hardware companies eager to build Grid products and to understand the
opportunities. Beyond the software and hardware companies we have
participation from telecommunications providers such as Level(3), Qwest and
BellSouth, as well.
But GGF participation isn't just the technology "suppliers" -- we have
"consumers" from industry as well. This year we have seen participants at GGF
meetings from dozens of end-user companies such as Johnson and Johnson,
Boeing, DaimlerChrysler, Merrill Lynch, Eli Lilly and Ford Motor Company.
Four years ago, when we founded GGF, there were only a small number of
commercial players involved, and all of them were technology companies. That
has grown steadily, but we also began seeing a rapid increase in commercial
participation and interest in mid 2002 as the Open Grid Services Architecture
(OGSA) effort began to converge. Earlier this year, we released the Open Grid
Services Infrastructure (OGSI) service specification. This enabling
specification was enthusiastically received and brought a new wave of
participation from commercial organizations, particularly from smaller
software companies and application developers.
The OGSA and OGSI working groups are co-chaired by leaders from industry
(Jeff
Nick from IBM, Hiro Kishimoto and David Snelling from Fujitsu) as well as from
research (Ian Foster and Steve Tuecke from Argonne National Laboratory and the
University of Chicago). Similarly, we have a strong set of activities
developing standards for integrating databases into open Grid systems, led by
both industry (Dave Pearson from Oracle) and research (Norman Paton from the
University of Manchester). An examination of the more than 45 working groups
and research groups in GGF will show that many of the most active groups
involve this partnership between commercial and science interests.
This is, in fact, the pattern we have seen in GGF -- the most successful
groups are co-chaired by a person from the research sector along side a person
from the commercial sector. We believe this partnering between commercial and
research leaders in an open, public forum is essential to developing useful,
widely accepted standards.
GGF is also more than just a community of technologists developing
specifications. We also have user community research groups oriented around
closely related applications. A year ago we formed the Life Sciences research
group which brings together bioinformatics, pharmaceutical and related
application end users. This year we have formed similar end-user application
groups around several scientific disciplines, including a social sciences
group. We are currently in the process of forming a similar group focused on
business and data center applications, which has come from discussions with
Fujitsu, Hitachi, NEC, IBM and HP. These groups are essential in informing the
standards process of the requirements of end users, and they serve as a
valuable mechanism for bringing commercial and science communities
together.
What about the different needs of science and research applications
relative
to commercial, business applications? Certainly, the applications are
different in many ways, just as they are in other contexts, but there are also
underlying standards that are shared. Commercial and science applications both
use the same Internet and Web technology, the same microprocessor and storage
technologies, and the same operating systems technologies. Commercial and
science Grid applications are also very similar. Whether your Grid environment
is intra-company or between many companies, there are fundamental security
technologies that are common. The differences are not in the underlying
standards but in the use of those standards within a particular context, and
those differences also exist in scientific Grids (e.g. within a single
university or laboratory).
It was stated in one of the articles about the Oracle announcement that GGF
releases versions of software that are not backward compatible. GGF doesn't
release software, but that's a fair criticism of early Grid tool developers.
It's also inevitable in the early stages of software development, and one of
the reasons that both research and commercial participants are investing their
time and effort in developing robust standards that can carry the technology
forward. It's important to separate criticisms of early software experiments,
or applications of standards, from the development of the standards
themselves.
There is a higher tolerance for failure in science and research
applications
than in commercial applications. This is precisely the reason that new
architectures and technologies have often been "proven" first in the research
realm and then later adopted by commercial applications. The Internet,
parallel computers, clustered computing, the Web and Web browsers, and Grid
technology are all examples of breakthrough technologies that have begun in
the research environment, moved to open standards and then applied to business
applications. That is precisely what we are seeing with Grid standards and
GGF, and it's been happening on a very fast timescale.
The nearly 50 companies sponsoring GGF and at least 50 more who are
actively
engaged, are working together to develop Grid standards that will create a
marketplace for Grid-enabled commercial business applications. My hope is that
Oracle will help us steer this exciting effort, and it seems to me that
creating a new consortium is something that runs the risk of slowing this
momentum rather than channeling it. I'm looking forward to the discussion!
|