Recently an article came out stating that APM was going the way of the space shuttle. Too expensive, outdated, not needed and so on. There were statistics given indicating that APM products were growing out of favor as well. Balderdash!
To my experience most customers don’t even use the acronym APM or what it stands for, application performance monitoring. Many enterprises have shared services, IT operations and application development groups and they all monitor or test the performance of their respective resources.
Statistics are a funny thing. Depending on when you ask someone their opinion, depends on what answer you will get. For example, someone asks you if you like a certain magazine. If the economy is good, your day is going well, no server problems, coffee is good, light traffic, you are going to say that the magazine is insightful, informative and you enjoy the read. But, the economy is not doing so well, there have been layoffs, the server just crashed and you got in an accident on the way to work, your response is going to be less than stellar.
The economy has had a huge impact on everyone in one way or another. When a person is laid off, another worker must assume the work and responsibilities left by the vacant position. If the person assuming the responsibilities has not been trained to use the installed products, the effectiveness of the products is not used to its full potential. Sometimes, the APM product is complex and being short staffed, there may not be time to train the new person. Correctly trained personnel will actually save you time and money.
APM products are here to stay. An un-monitored system? I shudder at the thought, as should you. The chaos and lack of production would be at an astronomical level with system crashes, memory leaks and failed applications. Cost and complexity need not be a concern with the right vendor.
Unlike what the author of this article expressed, it is certainly worth the effort for them to do this. We have customers transacting millions of dollars via their applications and you can be sure they are definitely monitoring them.
I think the real challenge is to ensure we are not fixated just on the data acquisition part of monitoring and are focusing on real-time analysis of the events we collect from monitoring so action can be taken before there is impact. This can’t be done effectively using eyes-on-screen monitoring. It requires analytic tools such as complex event processing to identify the application performance patterns that portray problem situations and initiate early-warning alerts.
For further insight into what defines a successful APM deployments, take a look at this list of APM case studies.