Research Note Decision Framework 05 August 2002

Gartner estimates that 40 percent of unplanned downtime results from application failures (see "Making Smart Investments to Reduce Downtime"). These failures can be caused by programming bugs, performance issues or changes to applications that result in breakage. Although many vendors offer solutions to provide Java application instrumentation after the fact (that is, in production), the best time to do this is during the initial development process (see "Application Server Management: Availability/Performance").

Management Time

A common complaint within the application development (AD) community is that there often is little time to add management functionality when the line-of-business executive team is pushing for the application to go "live." This results in the "robbing Peter to pay Paul" principle: The application goes into production; however, when problems develop (and they will), the cost in time and customer satisfaction to debug the application most likely would have paid for the extra few days' work to add built-in manageability. We estimate that for every phase in which a defect is left in beyond the original point of detection, it becomes 80 to 1,000 times more expensive to remove. This is certainly not a new phenomenon. However, the increasing visibility to customers of these applications raises the financial stakes.

It seems contradictory, but by taking time to add manageability, enterprises improve the time to market for this environment, because problems found in quality assurance and testing, not to mention development, will be easier to diagnose and repair.

This enables the application to be deployed more quickly. The benefits don't stop there. The IT operations group also should emphasize the much-reduced time needed for the developer to solve production-phase problems. Therefore, more development time can be spent on applications that add value to the enterprise, instead of tracing and debugging faults.

Although negative reinforcement is rarely recommended, the final weapon in the IT operations group's arsenal is to support lowered service-level objectives for nonmanaged vs. managed applications. Although acknowledging these reductions may be politically difficult to enforce in many enterprises, it is better to take a strong stand early. The finger-pointing will most assuredly start with the IT operations group when problems develop. As one Gartner client has remarked, the application server operations group has taken over from the networking organization as the place where interrogations now begin.

Process Produces Performance

The management of application server technology will impose new demands on the IT operations group and the AD organization. To optimize this complex environment, contributions will have to come from both organizations. However, it all begins in the design phase. Applications must be designed from the beginning with management built-in. This should not be an exercise solely limited to the participation of application developers. The IT operations group must provide input regarding specific needs for health and performance information — that is, the data that must be collected, its format, and where it should be stored or accessed. Effective IT operations groups will take the time to go beyond mere documentation in this area. They also will provide templates to the AD organization that can be incorporated into the code, with this information already embedded, which can be easily imported into an integrated development environment.

AD organizations, however, can't simply hand over the completed applications package to the IT operations group. As part of a formal application server production acceptance process, a run book that details operational support requirements also must be provided. This document should explain, in precise detail, component and scheduling dependencies, Java Archive (JAR), Enterprise Archive (EAR) and Web Archive (WAR) deployment needs, acceptable command syntax, and error codes and their meanings. It also should detail the information that has to be collected by the IT operations group, should the AD organization be asked to help with problem resolution. Test scripts developed in earlier phases also must be included in this package so that they can be implemented within a production-based monitoring system. Gartner believes that, over time, this informal alliance between the IT operations group and the AD organization will give way to a more-structured environment — the application server competency center (see "Application Server Management: Future Directions").

Management for the Masses

The good news is that building management into applications is not as difficult as it once was. For application server environments, the Java Management Extensions (JMX) specification is available for adding manageability to Java-based application servers. JMX offloads much of the burden of application developers having to understand the minutiae of management technology via the use of familiar Java semantics (see "JMX: The Key to Managing Java Applications").

For an even less-demanding technical approach, AdventNet offers ManageEngine, which enables developers to add JMX manageability to Java applications, including servlets and Enterprise JavaBeans. This capability can record, for example, the start and end times of specific methods and the number of times it is invoked. Conditions can be added so that alerts are generated based on desired thresholds. ManageEngine also includes a logging monitor Manage Bean that can issue events when certain messages get logged. ManageEngine is a stand-alone product; however, AdventNet plans to include its functionality as a plug-in for products such as Borland's JBuilder, which should further ease the application developer's burden.

Bottom Line: AD that considers application server management is no longer an option — the cost of online failures is rising every year. Enterprises must approach management from a multi-organizational perspective, ensuring that the appropriate level of instrumentation is embedded to maximize application server runtime performance and availability.

This research is part of a set of related research pieces. See AV-17-1684 for an overview.

Entire contents © 2002 Gartner, Inc. All rights reserved. Reproduction of this publication in any form without prior written permission is forbidden. The information contained herein has been obtained from sources believed to be reliable. Gartner disclaims all warranties as to the accuracy, completeness or adequacy of such information. Gartner shall have no liability for errors, omissions or inadequacies in the information contained herein or for interpretations thereof. The reader assumes sole responsibility for the selection of these materials to achieve its intended results. The opinions expressed herein are subject to change without notice.

Entire contents © 2002 Gartner, Inc. All rights reserved. Reproduction of this publication in any form without prior written permission is forbidden. The information contained herein has been obtained from sources believed to be reliable. Gartner disclaims all warranties as to the accuracy, completeness or adequacy of such information. Gartner shall have no liability for errors, omissions or inadequacies in the information contained herein or for interpretations thereof. The reader assumes sole responsibility for the selection of these materials to achieve its intended results. The opinions expressed herein are subject to change without notice.