March 17, 2010, 11:23 AM — How often have you heard the adage, "IT must align itself to the business needs"? It's one of those buzz phrases that sounds really obvious but, in reality, might be harder to achieve than one would think. Here is some insight on a tangible way that IT operations can absolutely add value to mission-critical business applications through a new generation of application performance management (APM) technology.
In the past five or so years, there has been a radical shift in the way that many business applications are built. Today's applications are fundamentally different -- built to take advantage of service-oriented architectures, open source software, agile development methods and new development frameworks. At the same time, the IT operations environment has gone from a physical data center to virtual servers and cloud computing; from building for peak demand to delivering capacity on demand; from developer-centric troubleshooting to operations-centric troubleshooting.
At one time, business applications were monolithic, with all the code executing within one big application server and the data being housed in one big database. If something went wrong with the application, one didn't have to look far to find the issue. That seems almost quaint today, when applications are designed to run over the Internet, and a business transaction may take dozens of hops to complete. Diagnosing a problem in an application today means having to follow a transaction all the way through its various processes taking place on many different devices.
In order to monitor the health of the business transaction, you need to understand what part of the transaction takes place where, and what impact that component's performance has on the transaction. This goes far beyond knowing if a piece of hardware is functioning properly; it includes following the flow of the business transaction across multiple tiers, understanding how long it spends in area of the application, and whether its journey meets the application's performance baseline based on historical patterns.
You can't get this level of insight from network packet inspection. Network tools don't support the full details of the business transaction because they don't go into the application code; they aren't "application aware." Since you can't manage what you can't see, it takes a different kind of tool to monitor today's distributed application.