IBM Adds ‘Traffic Cop’ Feature to WebSphere

IBM Monday unveiled a new grid computing software feature to its WebSphere application server that
automatically monitors application workloads and routes traffic from one
server to another in order to balance their workloads.

In what the Armonk, N.Y. company has described as a “traffic cop-like”
product, the feature allows a cluster consisting of many servers — from
dozens to hundreds — to operate as a single environment that adapts to
changes on the fly akin to grid systems. This is akin to grid computing,
where disparate computers and systems can become one integrated computing
system.

Big Blue said this feature can improve network performance by avoiding
missed server connections. The software is a key piece in what analysts
expect to see many such features from IBM in the wake of its company-wide
e-business in-demand strategy. Rivals such as Sun Microsystems, HP,
Microsoft, Veritas and Computer Associates have embarked on similar
strategies.

IBM added the feature to WebSphere because it said current server clusters
in data centers often do not handle unexpected workload changes. Because the
IT managers need to avoid network bottlenecks and keep users connected, they
usually plan for great spikes in demand but realize the potential of their
servers aren’t reached.

This goes to the heart of a major concern of enterprises needing IT
infrastructure today — they want to bolster return-on-investment (ROI) with
whatever equipment they do own or plan to buy.

Dan Powers, vice president of grid computing strategy at IBM, said customers
had asked for the ability to balance WebSphere workloads in a dynamic
fashion, so that’s what IBM provided.

Powers said one example where the “traffic cop tool” might be useful is for
a financial company that offers online trading. The company might have risk
management software running in parallel fashion. With their business policy,
they could take on more workloads when it isn’t busy with online trading.

Summit Strategies analyst Mary Johnston Turner discussed the import of the
new software tool from IBM.

“The availability of grid computing on the WebSphere platform is the first
tangible product to deliver on IBM’s On Demand Operating Environment (ODOE)
strategy,” Turner told internetnews.com. “ODOE is a multi-year effort
to tie all IBM products together in a highly dynamic autonomic operating
environment where many decisions about provisioning, workload balancing and
resource prioritization across multiple servers and systems can be executed
automatically, on the fly, in response to changing business requirements.
The WebSphere announcement is a down payment on this ambitious ODOE
promise.”

Turner said IBM has been one of the firms leading the charge on moving grid
into mainstream commercial applications and IT management solutions.

“But the next 6-12 months will tell us whether anyone outside the high
performance technical computing market is following,” she said. “When we see
major commercial ISVs and third party hardware companies announcing their
support for grid we’ll have proof that IBM is on the right track in betting
on grid’s ability to serve the broader commercial market.”

More broadly, Powers said features like the grid computing tool come from
IBM’s mainframe experience, where software features of mainframe operating
systems help the machines operate as much as 70 percent of their total
capacity. He also said customers can expect to see more features added to
improve WebSphere’s functionality down the road, including more parallel
processes.

Future versions of IBM WebSphere will extend the “traffic cop” capability to
disparate parts of a company and coordinate clusters of servers running
heterogeneous business applications, rather than just single clusters
running a particular application such as online trading.

WebSphere Application Server v5.0.2 enterprise edition is available July 25
for $30,000 per processor, and includes one free year of maintenance and
services.

Separately, IBM Monday said it has made some improvements to its iSeries
line of eServers, including a 5/8-way option with On/Off Capacity on Demand
to the iSeries 870. The new option geared for customers adding transaction
workloads or upgrading from AS/400 740 or i830 servers.

In conjunction with that, Big Blue also added two new iSeries servers built
for on demand environments: the iSeries for Capacity BackUp and the iSeries
for High Availability. The former is designed for companies requiring an
off-site, stand-by server for disaster recovery. The latter is targeted at
companies that require always-on high availability clustering.