Finding root cause of poor application performance

In my experience (in other words just ignore me :-) )  I have found that some coorelation can be accomplished via alerting and using different components in different applications.  But, we seem to miss more than the automation/coorelation picks up.

BMC makes a competing tool to Orion that they kindof claim will coorelate in the manner you would like to see.  But I have yet to speak with anyone that will rely on it.  I still say there is no substitute for a NOC/Server tech watching the screen.  A human can coorelate SO much more that our rules can.

In my opinion automation can only go so far, and humans must figure the rest out.

I am allways looking for better automation and smarter coorelation, so I'll be watching this thread closely.

There isn't a good way to do have the tools provide root cause analysis because that isn't what the tools are really designed to do.  I have worked with a lot of monitoring systems and none of them do root cause analysis, that is work that needs to be done by people.  People make the best correlation engines assuming they understand the infrastructure that they are working with.  A good monitoring system will provide you with that single pane of glass to see all aspects of your infrastructure so you can see all of the symptoms of the problem in once place versus having to dig around in a bunch of different tools with potentially conflicting data.

I like to compare it to the healthcare industry...

The different tools that the doctors use are like the monitoring system, they help expose the symptoms of the problem; however, in the end it's the doctor that makes the diagnosis, not the tools.