Consider this all-too-common scenario in today's complex, networked application environments. A network operations manager is awakened by a phone call in the middle of the night: "We're experiencing significant performance problems and we don't know why. We're losing users. They're frustrated and leaving the site without completing what they're doing."
The first thing everyone blames is the network, because the application is a networked application and network performance can be unpredictable. That means the first support call is to the network manager, who must investigate.
The diligent network manager immediately checks the network status to find out what the problem is. He or she is relieved -- but also frustrated -- when they find that the network status is perfectly fine and that there are no reported network problems. The network is operating well within its SLA.
Yet the performance problems exist. And then the finger-pointing begins:
- It's the application's fault.
- No, the developers left bugs in the code.
- No, it's QA's fault. They didn't find the problem during testing.
- No, it's the networks fault. The network management isn't reporting accurate results.
- And so on.
Left unresolved, the finger-pointing can create an unproductive communication divide between network operations and development teams. Communication seems the obvious solution, but the real crux of the problem is a lack of insight into the application tiers and the inability to identify where problems really exist.
|Editor's Note: The author, Tom Fisher, is a Senior Product Marketing Manager with Compuware APM dynaTrace. We have selected this article for publication because we believe it to have objective technical merit.
Enter the Next Generation of APM
For years, network managers have had proactive network management tools for managing network performance and diagnosing problems. They're usually able to eliminate the network as the source of the problem or triage problems quickly. But not until recently with the advent of a new generation of Application Performance Management (APM) solutions, such as Compuware dynaTrace, Quest Software's Performasure, New Relic, and AppDynamics, did application managers have comprehensive performance management capabilities. These solutions provide deeper and substantially stronger insight into application performance problems than that received by network managers.
In addition to problem diagnosis, DynaTrace, for example, supports end-to-end transaction path tracing from user clicks through the messaging layers and into the application servers and databases. It enables network managers to discover application performance issues at any layer of an application's architecture, whether they're source code, database SQL, messaging, network performance, third-party component or other problems.
DynaTrace also presents the transaction path performance information on dashboards ranging from simplified, business-aware dashboards down through complete transaction flow diagrams, UML sequence diagrams. It also supplies method source code in software development frameworks such as Eclipse. These business-aware dashboards, such as those for SLA management and transaction flow diagrams, can be combined with the network management information to enable network managers to monitor network and information flow performance through an entire application infrastructure. When an issue is detected, the location is immediately identified, enabling operators to drill down to the individual resources causing the problem.
APM Bridges the NetOps / Dev Team Divide
Network managers can then pass this accurate, detailed information back to the application developers in a format that the developers need and understand. This 'lingua franca' of actionable application performance problem information bridges the divide between application development teams and network operations and helps eliminate the finger pointing.
The key to bridging the divide between development teams and network operations is indeed communication -- with meaningful information in a common, understandable format. The new APM solutions provide that information in a format acceptable to application developers and network operations managers alike. Finger-pointing is eliminated and collaborative problem solving ensues.