Knowledge is power

Of course the first step in handling anomalous situations is to be aware that something is amiss. As proponents of the Unix philosophy have famously stated, a system must not just function well, but it must be seen to function well. This is called the rule of transparency, and in essence it admonishes application designers to build systems that provide visibility into their inner workings.

Taking this a step further, we might add that we should be able to know that the system is working even when we aren't looking. There are times when you may be actively watching the cluster, for example, when adding or removing nodes or deploying a new application. However, more often than not, you will have your attention turned elsewhere when the unexpected occurs. It is during these periods that you will need to rely on automated monitoring to alert you that there is trouble.

Monitoring via Java Management Extensions

Fortunately, Cassandra makes this easy by providing numerous Java Management...