|
 Grid
Computing projects have traditionally been associated
with Scientific Research. But grids can also add value
to business. Michael Monty, Business Development Executive,
Grid Computing Business, IBM (ASEAN/South Asia & ANZ)
brings us up to date with developments on the grid computing
front. By Brian Pereira
What
business value do grids and grid applications offer?
Grids can give business a competitive advantage and
reduce the time to market. For example designers in
a manufacturing company can work collaboratively on
a design in real-time, even though they are at separate
geographic locations. This enables what we call the
Virtual Corporation. It speeds up the design process,
for instance, and people need not come to a common drawing
board. You are setting up a virtual design environment.
The other area is the financial analytics space. Then
there is potential in areas like business intelligence
and data mining. Organizations today have vast amount
of data. The key focus in the industry is knowledge
management, and there's plenty of software for this.
The question is how do you analyze this data with the
computing resources you have? Most people can't, unless
they use supercomputers. The grid computing initiative
offers a way to utilize existing resources that we have,
which are underutilized.
The
other point is that there are so many variables involved.
There are a number of databases involved. Take the example
of government. Today there's a big focus on permanent
security. When you are checking a person's background,
you need to verify fingerprints, passports, birth certificates—across
different ministries and institutions. A grid offers
the technology to extract information from different
sources that may be on different computing platforms.
These various institutions need not modify their infrastructure,
as grids work in heterogeneous environments.
Can you cite some commercial applications for grid computing?
How are these being used worldwide?
In the oil industry you have players like Landmark and
Geoquest for seismic and reservoir analysis. We have
a project in Venezuela for a company that is working
on improving the quality of data. They are into oil
exploration and use supercomputing clusters. We are
seeing lots of interest in the Oil and Gas industry
for improving the quality of data through the use of
data grids and compute grids.
Would you like to comment on software development for
grid apps in business?
There are a number of companies working on grid software,
and IBM is working closely with five such companies:
Platform Computing, DataSynapse, Avaki, Entropia and
UnitedDevices. Even Sun Microsystems has its own grid
engine. Each of these companies focuses on a different
aspect of grid computing.
Within IBM we use our existing technologies like DB2.
With DB2 we have a federated tool, which means we can
put together multiple sources of data even from competitive
products (for data grids). We also have a tool called
DiscoveryLink within DB2 that discovers the data resources
within the grid, and it can pull back the information
that's relevant to the user query.
How does Grid Computing differ from Distributed Computing?
Where do Web services fit in?
Grid technology comes from distributed computing, which
has been around for some time. In fact Grid is the next
generation of distributed computing. Distributed computing
works in a homogeneous environment; grid computing is
for heterogeneous environments.
OGSA (Open Grid Services Architecture) and Web services
are intrinsically linked. A lot of what the OGSA standards
promote are, in fact Web services. We are talking about
open standards: XML, Web services description language,
SOAP etc. Web services look at application services.
It looks at how well I integrate multiple applications.
Grid services look at how I design grid middleware so
that they can all work together. The grid middleware
itself are applications and they have to communicate
using Web services.
What are IBM's grid computing initiatives? What are
the key focus areas?
IBM's grid offerings combine its existing services,
software and hardware, with middleware from other companies
such as DataSynapse, Avaki, Entropia, UnitedDevices
and Platform Computing. The solutions are designed specifically
for R&D, Engineering & Design, Business Analytics,
Enterprise Optimization, and Government Development.
In all, IBM has 10 industry specific grid offerings.
Internally, we have written a number of whitepapers
and a Red Book on grid computing. We are educating customers
too. IBM has also set up three e-business on demand
centers around the world. Within these centers we have
grid briefing centers.
In addition, IBM is engaged with the top 20 financial
companies on grid computing projects. Some of these
are brokerage Charles Schwab and J.P. Morgan. Through
grids we are aiming to reduce the time cycle of financial
analysis applications by 90 percent. This will enable
companies to do calculation throughout the day instead
of end of day.
Are there any issues that need to be resolved before
grid computing takes off?
Grid computing in its purest form presents a virtual
environment. There is a long way to go before we get
to that. But there are specific projects for it now.
The biggest challenge that I see is a Soft challenge.
I don't see a problem with technology because in the
long run, technology evolves quickly.
The Soft challenge lies within organizations and government
institutions. People working in different departments
are averse to the idea of sharing computing resources.
Research bodies don't cooperate with each other. Some
universities don't like each other.
You need to have a top down approach for grid projects.
You need to have someone with authority insisting that
everyone works together in the interest of the nation
(or organization). In Singapore and the UK, they have
national grid computing projects that are supervised
by high-level steering committees. They have government
servants on these steering committees. The other problem
to address is security. But it's not really a grid issue,
it's a technology issue.
|
What
is Grid Computing?
Grid computing links several computing resources
over a network, to share data and computing resources
(processors, storage etc). Since grids support
heterogeneous environments, the computing resources
on a grid can be on different platforms, and are
usually in different geographical locations.
The combined processing power of all computers
in a Compute Grid can be equivalent to (or exceed)
that of a single supercomputer. So Compute Grids
offer (research) institutions tremendous computing
power, without the need to invest in an expensive
supercomputer.
There are also Data Grids that link disparate
sources of data (in various databases). So, with
a single query, a user can pull information from
several databases that may be at different geographical
locations, and on different computing platforms.
So far, it was mainly the scientific & research
institutions that leveraged on the potential of
grid computing. Grids have also been used for
oil exploration, and in the aerospace industry,
for design.
The next area for Grid usage is business. Grid
computing in business will take off once suitable
applications (middleware) become available, and
when businesses realize the benefits of sharing
computing resources on a grid.
|
|
Standards
The Globus Project (www.globus.org) and the Global
Grid Forum (www.gridforum.org) are two major worldwide
initiatives towards establishing common standards
for grid computing. Then there are emerging grid
computing standards like OGSA.
The Globus Project is an open source initiative.
It is developing the fundamental technologies
for building computational grids. It offers the
Globus Toolkit (now in version 2.2), an open architecture,
open source software toolkit that enables organizations
to build computational grids that support their
applications.
OGSA (Open Grid Services Architecture) is a proposed
evolution of the current Globus Toolkit towards
a grid system architecture, based on an integration
of Grid and Web services concepts and technologies.
The Global Grid Forum (GGF) is a forum of over
5000 individual researchers and practitioners
working on distributed computing or grid technologies.
|
Brian
Pereira can be reached at brianp@networkmagazineindia.com
|
|| Feedback
|| Subscribe-
