THE EVOLUTION OF GRID STANDARDS 12.02.02
Andrew Grimshaw, CTO and Founder, Avaki Corporation GRIDtoday
==============================================================================
Andrew S. Grimshaw is Professor of Computer Science at the University of
Virginia and founder and CTO of Avaki Corporation. His research interests
include grid computing, high-performance parallel computing, heterogeneous
parallel computing, operating systems, and high-performance parallel I/O.
He is the chief designer and architect of Mentat, Legion, and Avaki.
Grimshaw received his M.S. and Ph.D. from the University of Illinois at
Urbana-Champaign in 1986 and 1988 respectively.
In this series of articles, Dr. Grimshaw will discuss the state of the art
in grid computing, including the definition and benefits of grid computing,
how grid software provides a return on investment, and this issue's topic:
standards efforts in grid technology.
Open Grid Services Infrastructure (OGSI) [1] and Open Grid Services
Architecture OGSA [1] are standards emerging from the Global Grid Forum (GGF)
http://www.gridforum.org .
These are today's prevailing standards initiatives in grid computing, and
will, in my opinion, define the future of grid computing.
This article explains why grid standards are needed and outlines the current
state of the OGSI and OGSA standards.
Why Grid Standards Are Needed?
In grid computing there is a strong need for a widely agreed upon, well-
designed and well thought-out architecture to tie the grid together. Without a
common architecture and interoperability protocol, different vendor solutions
will not talk to each other, and will form disjoint islands of grids.
Grids stitch together and make available, in a uniform, secure way, many
heterogeneous resources that may reside in different locations or
organizations. If different sites or enterprises select different grid
software, and that software cannot interoperate, then the overall value of the
grid to the enterprise is reduced. It is this value that will drive the
widespread adoption of grids. (More on the value proposition and the ROI case
for Grids in a later article.).
Like any standardization process, standards will impact grid computing in a
practical way if they are widely adopted and facilitate interoperation,
customer flexibility, and choice. Consider two different grids that might be
represented by OGSA-defined resource management services that represent
computational resources. These two resources would be accessible (subject to
access control and policy) by a high-level job placement tool that
"understands" OGSA resource management services and can send a job to either
of the two grids.
Further, because Grid Services are based on Web Services, the widespread
adoption of OGSI/A by the grid community will ease the transition of grid
technology into mainstream enterprise computing.
Why it is the right time for standards?
Some argue that it is premature to standardize in the grid community, and that
there has been a lack of experience and consensus building such as was the
case, for example, in POSIX Unix.
I feel that this is not the case and that now is the time to standardize.
There are three reasons.
-- First, there is in fact a long history and rich literature in grid
computing and its direct "ancestor" distributed systems. The first grid
projects began almost a decade ago (called meta systems then), and many of the
most important issues in grid computing (and addressed in OGSI/A) such as
naming and binding, are really distributed systems issues about which much is
already known. The community has had a long time to experiment with and digest
this work.
-- Second, there is a compelling need: if we do not standardize we will suffer
from balkanization and run the risk of failing to deliver the promised value
to organizations in both public and private sectors that will sustain the
growth and development of grids.
-- Third, both the long-term stake-holders in grid computing and the newcomers
are ready and willing to make the compromises necessary for progress and to
come together now. In a few years the cost for any particular stakeholder to
adapt to a new standard will be prohibitively large making consensus on a
usable standard very difficult.
The Content of OGSI and OGSA
OGSI extends Web Services via the definition of Grid Services.
Mechanically the differences between Grid Services (OGSI) [1] and Web Services
are few: A Grid Service is simply a Web Service that conforms to a
particular set of "conventions." For example, Grid Services are defined in
terms of standard WSDL with minor extensions, and exploit standard Web Service
binding technologies such as SOAP, WS-Security, etc.
However, it is this set of "conventions" that also fundamentally sets Grid
Services apart from Web Services.
Grid Services introduce three fundamental extensions: § Named service
instances, and a two-level naming scheme that facilitates traditional
distributed systems transparencies § Services that have a minimum set of
capabilities including discovery (reflection) services; and § Explicitly
stateful services with lifetime management. OGSI is being developed in the
OGSI-WG (working group) in the GGF.
The form of the proposed OGSI standard is fairly well known, and the
specification is expected to emerge from the standards process in the second
quarter of 2003. OGSI will be the basic interoperability layer (in terms of
RPC, discovery, etc.) for a rich set of higher level services and capabilities
that are collectively known as OGSA, the Open Grid Services Architecture.
OGSA is being developed in the OGSA-WG of the GGF. The OGSA-WG is an umbrella
working group within the GGF that will spin off more specialized working
groups to develop specific service standards that, together, will realize a
"meta-operating system" environment.
The first of these working groups formed is the OGSA-Security working group.
Working groups on topics such as "resources" (hosts, storage, etc.),
scheduling, replication, logging, management interfaces, fault-tolerance, etc.
are anticipated.
Progress Of The Standards Initiatives
OGSI is expected to be available as a formal standard in the middle of 2003.
That said, there are already several projects, both in the government/academia
and in industry, to build services that will be rendered via OGSI as well as
implementations of OGSI.
For example, the Globus research group at Argonne National Labs and USC ISI
is actively developing a reference implementation of the OGSI standard in Java
(C bindings will be available as well. The Globus team has released early
versions of the Globus Toolkit that both use and implement OGSI.
Similarly Avaki has announced that it will also support access to Avaki Grids
via OGSI. The OGSA-WG was formed only recently, and its full scope has not
yet been worked out. It is reasonable to expect that it will be some time
before the high-level services and their inter-dependencies have been
identified and new working groups spun off.
Summary
OGSI and OGSA are the Grid Standards efforts today. They are being developed
in the GGF and are gathering momentum in the Grid community. Avaki is an
active participant in both efforts.
We expect that the Grid community as a whole, and grid users in particular,
will benefit tremendously from the adoption OGSI and OGSA by Grid vendors. The
needs for standardization in Grid computing are likely to change over time.
Higher level services and standards will be defined, while some lower-level
standards will need to be re-examined.
The adoption of standards serves a greater good: organizations' ability to
achieve the highest possible return on their investment in grid technology.
References [1] I. Foster, C. Kesselman, J. Nick, S. Tuecke, "The Physiology of
the Grid: An Open Grid Services Architecture for Distributed Systems
Integration."
Open Grid Services Infrastructure WG, Global Grid Forum, June 22, 2002. S.
Tuecke, K. Czajkowski, I. Foster, J. Frey, S. Graham, C. Kesselman, "Grid
Service Specification." Open Grid Services Infrastructure WG, Global Grid
Forum, Draft 3, July 17, 2002.
***************************************************************************
Full background information on all Sponsoring Companies
[ ] 921) SGI [ ] 934) Hewlett-Packard
[ ] 527) Intel [ ] 942) Sun Microsystems
[ ] 909) Fujitsu [ ] 943) Linux NetworX
For sponsorship information contact: gridads@gridtoday.com
GRIDtoday welcomes bylined comments for publication.
***************************************************************************
Copyright 2002 GRIDtoday Redistribution of this article is forbidden by
law without the express written consent of the publisher. For a
subscription to GRIDtoday, send e-mail to gridfree@gridtoday.com