IBM Accelerates WebSphere Application
Development Via Fast Diagnostics

Innovative software pays off—if it’s developed optimally

Cloud & smarter Infrastructure Weekly. An IBM service management perspective. If you want to stand out in a challenging business climate, software-driven innovation is a great way to go about it. Software developed in-house can be designed from the start to fulfill particular business goals or strategies, make optimal use of particular resources, and work within an existing infrastructure in ways that shrinkwrapped software can't.

Achieving those outcomes, however, does put the burden of software optimization on the organization in new ways. It's critical to ensure that applications are as feature-complete, performance-enhanced, and bug-free as possible before they're deployed into production; otherwise, the soon-to-be-discovered issues can easily lead to a worse outcome, not a better one. Specifically, the new application will have to be pulled from production systems and its shortcomings analyzed and resolved.

At this stage, reports on application performance are typically inadequate because while they quantify a performance shortcoming, they don’t identify its root cause. Instead of showing why an issue exists, in other words, the report usually only shows that a problem exists.

This situation leads in turn to multiple cycles in which a new build is created and tested, reports are generated, and problems are found... but solving those problems takes longer than it should. These delays ultimately slow the process of rolling the new application into production and receiving the intended value.

IBM Application Performance Diagnostics: Intuitive, comprehensive assessment of in-development WebSphere applications

In order to provide information in the most intuitive way for different circumstances, the solution presents analysis via a number of related formats in a Performance View. The idea is to give developers several different perspectives into application performance—each a big picture from which it is then possible to zoom in on particular information as needed.

For organizations in this situation with respect to Java applications running in IBM WebSphere environments, this problem is now a thing of the past.

IBM Application Performance Diagnostics Lite, just graduated from a continuous open beta into a formal release, provides an intuitive, lightweight, and surprisingly powerful way to assess new applications—in as much or as little detail as you need to solve discovered problems.

That means applications move into production more quickly, generate more value over a longer period of time, and are much less likely to be rolled back to development, all of which contribute to a better business outcome.

A lightweight design that delivers heavyweight results

In essence, the solution consists of two elements. The first, a data collector drawn from IBM Tivoli Composite Application Manager technology, is installed on an instance of the application server used for testing purposes. It collects diagnostic data (particularly regarding the application's response time to user requests) and provides it either in real time via streaming, or periodically as data files that can be analyzed at the engineer's convenience.

The second element is the Application Diagnostics Lite client, which can be installed and run on any convenient workstation. It provides analysis of the data, presentation of the results, and it allows the team member reviewing the results to drill down into the data as necessary to isolate the root cause of any discovered performance shortfalls.

Because the client's own requirements are modest, it can be installed wherever it's needed; all that's required is the Eclipse environment (itself included with one version of the client). And assuming the data is collected in a file (rather than via live streaming), analysis can also happen anywhere, at any time—whether the workstation is connected to the application server instance or not.

Assess application performance in the level of abstraction, and amount of detail, you need

In order to provide information in the most intuitive way for different circumstances, the solution presents analysis via a number of related formats in a Performance View.

The idea is to give developers several different perspectives into application performance—each a big picture from which it is then possible to zoom in on particular information as needed. These perspectives include:

Finally, beyond the Performance View, there is additionally a Profiling View. This depicts aggregate statistics information for all the class name and method name for all the method calls being made—a coding-centric perspective that can help developers determine cases in which a particular call is underperforming against expectations.

The result? Thanks to the extensive analysis, and multiple perspectives on it, development teams can quickly find problem areas in code, isolate them to root causes, and address those causes—so applications can be finalized and rolled into production faster, while the odds of a problematic release will significantly decline.

Instead, the application simply creates the intended value by supporting (or providing) the necessary service, and the organization is rendered that much more competitive in the marketplace.

Additional information

Recent Articles

Contact IBM

Considering a purchase?

Stay Ahead at Innovate 2013

Innovate 2013

With over 450 outcome-driven sessions, a cutting-edge Expo, labs, certifications and evening entertainment including a night at Disney's Animal Kingdom, this five-day event promises to be THE technical education event of the year for software and product practitioners.