As Web-based applications proliferate and gain mission-critical status, a clutch of companies are vying to provide new types of tools that better monitor and manage application performance across a distributed infrastructure.
This week, companies such as ProactiveNet Inc., Mercury Interactive Corp., and startup Panacya Inc., will roll out products that take a holistic approach to pinpointing the source of bottlenecks and coding glitches that can lead to application slowdowns.
These newer players are moving away from traditionally reactive methods and embracing such things as automatic baselining or "self-learning" of optimal application performance, as well as correlation and root-cause tools that monitor attributes across the entire software infrastructure. In many cases a problem lies not in the app itself, but somewhere else in the enterprise food chain.
By eliminating exhaustive sifting of individual application data that can lead to a diagnostic dead end, these new tools aim to cut down on IT resource requirements that drive up costs. In addition, they will work in concert with existing network management software, and represent another step forward in the quest for "self-healing" systems such as engendered in IBM Corp.'s Project Eliza hardware initiative, said industry analysts.
"We aren't there yet, but eventually auto-baselining will merge with self-learning and then be directly tied to self-healing," said Corey Ferengul, an industry analyst at Meta Group Inc.
Santa Clara, Calif.-based ProactiveNet, which counts eBay and Wal-Mart among its customers, is unveiling a version of its management software that is aimed at application servers, including BEA's WebLogic, IBM's WebSphere and Sun Microsystems' Sun ONE (Open Net Environment). The core software uses intelligent thresholding and root-cause analysis to monitor the health of a transaction in the context of the entire application delivery path, said Ajay Singh, ProactiveNet CEO.
"Philosophically, most solutions focus on availability -- you're either up or down," Singh said. "But with Web applications, speed and response are key factors. Being slow is just as negative as being down."
McKesson, a San Francisco-based supplier of healthcare products, is using ProactiveNet for Application Servers to track the availability and performance of its supply management online system, which is the core application behind its order management services. The company preferred the ProactiveNet product over several best-of-breed tools because of its broader range of infrastructure coverage, according to Morgan Morris, vice president of architecture at McKesson.
ProactiveNet for Application Servers monitors all the touchpoints for the app server, including Web servers and databases. It is priced starting at US$300 for two-CPU servers.
In its company launch this week, Panacya will unveil its bAware application performance management suite and is expected to ink a deal to manage BEA's myBEA program applications.
Panacya's bAware software suite offers a semantic modeling studio, real-time monitoring, peer-to-peer analysis, and correlation tools, according to Franco Negri, CTO of the Annapolis Junction, Md.-based company. Panacya relies on its Data Provider agents to extract application information and Data Provider Server, a secure object access protocol (SOAP)-based Web service for standards integration.
Meanwhile, Mercury Interactive this week will launch ProTune, a software version of its ActiveTune service that enables customers to "tune" applications in a production environment to understand their behavior. The software measures impact upon server clusters, databases, load balancers, and transactions, said David Gehringer, director of product marketing for tuning solutions, at Sunnyvale, Calif.-based Mercury.
"One of the challenges today is the whole concept of distributed Web services. How do I know this application is written is really going to work? You can't test that relationship in a lab, it isn't going to work," Gehringer said.
After taking a baseline measurement of a production system, ProTune identifies and offers recommendations for fixing isolated bottlenecks, Gehringer added.
Mercury and the others will find themselves up against industry heavyweights in the "tug-of-war" over self-healing, and could be cannibalized by large equipment vendors that can offer autonomic computing in their hardware, according to industry analyst Paul Bugala of Framingham, Mass.-based IDC.
IBM, for example, recently released IBM Tivoli Monitoring 5.1, which includes an autonomic engine and best practices mechanism to "identify-notify-cure" ailments for systems and applications, said Chris O'Connor, IBM Tivoli director of performance and availability solutions.
"Many applications in today's e-business environment are n-tier applications that integrate components that were not designed to run 24x7," O'Connor said. "The first step towards self-healing is identification of problems, or pending problems."
Baselining serves as a simplified way to approximate normal conditions and extrapolate problems, said O'Connor, adding that IBM Server Group and Tivoli will ship pre-baselined hardware and software for optimum hardware utilization.
But O'Connor did point out that IBM's efforts are intended to also allow third-party network management vendors to plug into and extend its management solution beyond firewalls.