Monitoring and managing clustered servers

[wingkwan]

Looking for a way to monitor a clusterable enterprise server product.

Each individual server (node) is collecting individual counters and statistics. Need to provide a aggregation view across the cluster. Some of the counters could be the sum of all individual value. In the case of throughput the aggregation might be an average of all the individual servers throughputs.

Any suggestion will be appreciated.

Wing
[e-mail]

[brucespringfield]

How are you collecting individual counters and statistics?

[wingkwan]

Thinking to use AOP to store info in MBeans.

Or provide inline code to collect counters.

But the key question is how to aggregate counters across server nodes.

[brucespringfield]

You might consider setting up a JGroups Channel and use multicasting to pass information among the nodes as does the JBoss Cache. One idea.