Your day is going great.  Traffic was light, coffee is good this morning and the application release that you’ve been working on forever is being released today. One more acceptance test run and it’s a go!  Life is good….. until the call from server support. During final testing, the application started hoarding resources, otherwise known as a leak. Abnormal usage continues to rise.  Support had no idea that there was a problem. The server has failed.  Everyone from the CIO to the guy in the mailroom is in a panic.  This has happened more times than anyone wants to admit.

In most companies, you either have someone that sits and stares at a screen all day, in the hopes that maybe, just maybe, they might see something that might indicate that there is a problem.  Or, you have the support group that can’t react until the server fails and then they panic and make excuses as to what happened.  Neither of these situations is productive nor effective.  If only there were some way to be alerted that there is a failure just waiting to happen.

There is… consider monitoring applications for abnormal behavior and not just generating alarms when a singular metric exceeds a threshold.  Among the many things you can look at is Heap Utilization and a Momentum Oscillator.  They can be used together to identify when a resource is creating abnormal usage.  As the usage rises, warnings are sent to the support person so that adjustments can be made or the resource shut down.  Application leaks must always either be prevented or fixed.  The server is always busy, but by monitoring behavior more than individual metrics, you can be in better control of what your server is doing.  Is it going to handle the thousands of transactions required or is it going to be spending a lot of time doing garbage collection, performing poorly and restarting?  Another thing to consider is the scalability of your monitoring tooling and whether it can handle as many resources as you need monitored, regardless of the underlying platform or technology.  You can’t afford down time and by effectively monitoring behavior, you are heading off and preventing the possibility of a server failure.