In-Depth

The SOA Frontier: CA Troubleshoots Application Performance Problems

CA pushes Wily SOA Manager as a good prescription for treating application performance issues in SOA environments.

• By Stephen Swoyer
• 03/27/2007

If you build it, you’ll also have to manage it. That’s particularly true in the wild and unruly world of Web services, where an “application” is often no more than the sum of its composite services. SOA brings with it a whole new kind of application management headache—along with a whole new breed of application management tooling.

Computer Associates International Inc. (CA) last week introduced its first SOA management offering, Wily SOA Manager, an offshoot of its Wily Introscope application management tool. SOA application troubleshooting can be particularly troublesome with regard to composite applications, which (in some cases) are “applications” that consist entirely of services (e.g., an online shopping cart application that calls a CICS Web service and also consumes a predictive analytic service running on a distributed platform). The point, CA officials say, is that Wily SOA Manager can automatically identify dependencies among Web Services, and—once the discovery process is completed—can identify performance problems or bottlenecks as they occur. This can shorten downtime and—potentially—help companies either meet or enhance service level agreements.

“[T]he Web Services that an SOA enables are often reliant on the performance of other, underlying components,” said Mike Malloy, vice president and head of products at CA Wily, in a statement. “Because Web Services operate in the context of existing, heterogeneous applications, companies require management capabilities at every level of their application infrastructure. Wily SOA Manager meets this challenge by providing IT teams with a single lens that brings the performance of their entire enterprise into focus.”

Wily SOA Manager doesn’t just provide an abstract view of Web services, CA officials say; it also identifies the back-end systems that are tied to (or which expose) such services in the first place. It ships with pre-configured dashboards that provide operations and application support personnel with an at-a-glance view of service response times, call volumes, and errors in real time.

In addition, Wily SOA Manager provides error detection and impact analysis capabilities, which permits support personal to monitor application performance and verify the integrity of heterogeneous transactions.

Other features include synthetic transaction generation (which permits the monitoring of business process performance and availability via Service chains); service groups, which not only supports automatic discovery of service dependencies and data collection (from UDDI repositories), but also lets operations and production support staff group services together for more intuitive alerts and reports. There’s also a requisite reporting component, with capabilities that address not just capacity planning but compliance requirements, CA officials say.

Wily SOA Manager requires CA’s Wily Introscope. It supports most commercial application server platforms, including Apache Axis, BEA WebLogic Server, IBM WebSphere, SAP NetWeaver, and the Microsoft .NET Framework.

IBM officials say SOA—and the quasi-related uptake of WebSphere Application Server on z/OS—are two important contributors to the mainframe’s resurgence. Organizations tap SOA to expose mainframe assets—which must otherwise be yoked by means of cumbersome enterprise application integration (EAI), extraction transformation and loading (ETL), or enterprise information integration (EII) tools—to distributed environments.

Not only are most prominent EAI, ETL, and EII tools Web services-ready—such that an application can consume an IMS ETL feed that’s packaged (and appropriately transformed) as a Web service—but key mainframe applications, such as CICS itself, are also Web services-friendly. The result, IBM officials say, is that the mainframe isn’t so much a stalwart survivor of but a full-fledged player in the enterprise application-scape of today.

“A lot of customers want to get their heads around SOA, and if they have the mainframe [already] in their shop, that definitely works as the heart of my SOA infrastructure, of my enterprise service bus,” says Randy Daniel, director of System z marketing execution at IBM. “We’re hearing from a lot of mainframe customers who are planning [their SOA] efforts around their mainframe investments.”