 |
|
DAILY NEWS AND INFORMATION
FOR THE GLOBAL GRID COMMUNITY / JULY 28, 2003; VOL. 2 NO. 30
|
Special Features:
APPLICATION TRENDS IN VERTICAL
MARKETS PART II
by Insight Research
Commercial Applications
Within commercial enterprises, grid computing has found early adopters in
life
sciences, technology, and engineering companies. Many are motivated to
explore and use grid technology by its promise to reduce costs through more
efficient utilization of IT resources. In addition, the dramatic increases in
productivity that can result are very important to technology-based companies,
as they can translate to faster time-to-market for new products and services,
thus improving competitive position. Finally, and perhaps most significantly
for the longer term, the ability of grid technology to quickly enable "virtual
organizations" of workers to collaborate within and between organizations
across the world is also of increasing interest to enterprises.
Pharmaceutical, Biomedical, and Biotechnological Markets
Life sciences companies have many applications that are a good fit for grid
computing. Virtually all the life sciences are in the throes of becoming
intensively analytical with a strong information basis. Advanced research has
resulted in an explosion in the amount of data being generated.
In addition to having significant numbers of computing-intensive and data-
intensive applications, many life sciences projects are highly collaborative.
The requirement for collaboration among various departments and organizations,
as well as the competitive pressure for rapid success, drives the need for the
kind of "virtual organizations" that grid computing can provide. All of these
developments make life-sciences companies a good fit for both compute grids
and data grids solutions.
Pharmaceutical companies are among the leading early adopters of grid
technology. These companies all face the same dilemma of watching research
and development costs increase exponentially, while it becomes more time-
consuming and expensive to get drugs out of the production pipeline and
through FDA approval and market launch. Further squeezing the
pharmaceuticals' bottom line is the growth of buyers' purchasing power with
the spread of managed care, and this has put significant downward pressure on
prices. Many pharmaceuticals, both large, established organizations and
start-ups, are turning to more computational work, or "research and
development in silicon", as one way to lower costs. Drug discovery
simulations require supercomputing power to perform the intensive application
processing. Compute grids are ideal solutions for these applications. One of
the leading grid ISVs predicts that by 2004-2005 nearly 25 percent of research
will be done in silicon.
Engineering and Design Automation
Airplane part design - A New-England-based aeronautical engineering
division
of a multinational industrial conglomerate started using distributed computing
in early 1990's to model jet engines and gas turbines. The division had found
it too slow and expensive to rely solely on physical testing of these complex
types of equipment, and so they switched to large-scale numerical simulations.
The division uses commercial grid software on 5,000 workstations and 150
servers at five locations in the US and Canada. The simulations are executed
at night using the workstations' and servers' idle processing power. If not
enough processors are available at one location to execute a particular task,
the task is transferred to another location.
Computer chip design - The engineering division of a major electronics
manufacturer used grid technology to meet an accelerated deadline for
developing the company's latest chips for cellular telephones. Although the
engineers had already made significant progress on the design, fiscal
constraints effectively barred the division from purchasing any additional
compute resources to meet the new deadline. The division decided to install
software from one of the leading grid vendors to help them combine available
compute resources (about 400 workstations) into a virtual resource pool.
Within several weeks, the software was installed and configured, and the
engineering staff was trained to send their compute-intensive jobs to the
Grid, which then found the most appropriate systems to run the jobs
transparently.
Computer animation and video postproduction - A small animation company
specializes in digital special effects for commercials, television series, and
films. The digital special-effects business is host to a constant concern
with the time and effort it takes for animators and technical directors to
create many versions of various shots and farm these shots off to be rendered
frame-by-frame for review. Since "rendering" is such a compute-intensive
task, finding the available computing resources and distributing the work
effectively can end up being tremendously time consuming. After deploying
grid software within their server farm, animators could submit any process-
animation or vector rendering-with the same command.
Aerial and satellite image distribution - GlobeXplorer is a three-year old
California start-up that calls itself the leading provider of satellite images
and aerial photography via the Internet. GlobeXplorer claims to warehouse the
world's largest commercial database of aerial photos and satellites images,
which are supplied by a network of global content providers such as Earthsat,
Space Imaging, Worldsat, and others. To increase overall utilization of its
computing infrastructure, the company deployed an intra-company grid to allow
resource sharing across all computers using commercial grid software. As a
result, the company can handle three times the load with the same equipment.
It estimates it saved one million dollars in the first year of operation.
Financial Services
Investment banking applications - The investment banking unit of one of the
leading global financial services firms is introducing an innovative new
internal utility computing service for use in its compute-intensive trading
and risk management applications. Rather than dedicating IT resources to
specific lines of business and applications, the unit will provide its lines
of business with processing power for these applications from a combined
virtual pool of resources. The goal is to improve service levels and to cut
costs by charging the business units for the resources that they use, with
utility-style peak and off-peak pricing. The bank is working with one of the
leading grid ISVs to develop and deploy the kind of intelligent software
infrastructure needed to meet these objectives.
Life insurance financial modeling application - In early 2001, an analyst
at
a
leading life company was given the task to create a financial modeling
application to predict how different market scenarios would affect the
company's investments. He used a specialized mathematical application
development tool to complete the programming in a few weeks. Running the
application on his desktop PC, however, the one GHz Pentium III with 256MB of
RAM proved completely impractical. It took days to complete the thousands of
calculations necessary to analyze each scenario. The analyst's company had
already investigated the use of a grid software product and decided that the
analyst's application was a good place to try it out. As a result of the
successful test, the company will soon install a large PC grid using over 100
employee desktop PCs.
Risk management applications - A grid software company's first customer
moved
a system used for trading and managing of fixed-income derivative securities
into production powered by a local intra-enterprise grid. When traders,
portfolio managers, or middle office managers perform profit and loss
statements, request real-time decision support like pricing and hedging, or
perform regulatory risk reporting, the Grid powers the calculation,
distributing it among 200 processors. Other customers have deployed Grid for
similar trading floor applications for equity and credit derivatives
operations.
Consumer Applications
In May 2002, IBM and Butterfly.net announced that they were deploying the
first ever commercial grid service aimed at the fast-growing online video
gaming market. IBM's participation in the announcement is significant, in
that it represents an extension of grid technology beyond its traditional
roots in science and engineering into its second adoption phase within
commercial enterprises, and even into the third phase of consumer-based
applications. It is also significant in that it is an early example of a
business model that leverages grid technology based on services, in addition
to software.
West Virginia-based Butterfly.net is a small online game development
studio.
With years of experience in game development, the company has recently
extended into the realm of massively-multi-player games (MMGs) that run on
PCs, consoles, and handheld devices. MMGs are a rapidly emerging sector of
the video game market, in which millions of gamers worldwide play together in
real-time immersive 3D worlds.
Butterfly.net determined that grid technologies could help fulfill their
vision. Working with IBM Global Services over the past two years and using
the Globus Toolkit, the company has built customized gaming software and a
grid that distributes the processing of game interaction across a network of
server farms. Powered by IBM servers running the Linux operating system, the
Grid uses the Globus Toolkit to provide a resilient infrastructure where
computing resources can be allocated to the most popular games and busiest
areas, and servers can be added or replaced without interrupting game-play.
The server farms are hosted in IBM Global Services (IGS) data centers around
the US, with plans to extend to IBM centers worldwide. Butterfly.net calls
itself "the leading grid computing provider for the online video games
industry."
|