by Michael Procopio

In my IT days (it has been a while) as still happens today this is the question many have asked. It’s more complicated today, applications are more distributed now. However, you still have to go through the triage process. The topic these days is named “Application Performance Management” or APM.

APM has two parts, the traditional looking at infrastructure resources and measuring performance from the end user perspective.

You probably detected this problem in one of two ways. You are ahead of the curve and have end user monitoring in place or the user called the help desk to complain.

A typical web based application today uses a web server, application server and a backend, typically a database. Though the backend might actually have multiple parts if service oriented architecture, (SOA) is used. Good news, Operations Manager, agent based, and SiteScope, agentless, will provide status on condition of those servers.

These tools can also look at how packaged applications are doing on the server. Oracle, WebSphere, MS Exchange, MS Active Directory, to name a few, can be monitored by either Operations Agents or have SiteScope templates (a SiteScope template is a prepackaged set of monitors). These tools might point to something as detailed as database locks being far higher than normal and beyond the current setting on the database. A quick parameter change might fix this.

Next, we have the code. We hope this isn’t the case because this typically moves the problem from operations to development. However, Operations is still responsible for pinpointing the problem area. This is typically the domain of application support and in some organizations that’s inside Operations, in others a different group.

Here Business Transaction Management (BTM) tools can help. BTM manages from a transaction point of view. BTM includes transaction tracing. TransactionVision and Diagnostics work in a complimentary fashion to give you the next level of detail although each is usable separately. TransactionVision traces individual critical transactions (as you define them) through multiple servers; it gives you information on a specific transaction including the value of the transaction.

Diagnostics provides aggregate information on all transactions in a composite application giving you timing information. It can pinpoint:

· where time is spent in an application; either processing data or waiting for a response from another part of the application.

· the slowest layers.

· the slowest server requests which are the application entry points.

· outliers to help diagnose intermittent problems.

· threads that may be contributing to performance issues.

· memory problems and garbage collection issues.

· the fastest growing and largest size collections.

· leaking objects, object growth trends, object instance counts, and the byte size for objects.

· the slowest SQL query and report query information.

· exception counts and trace information which often go undetected.

TransactionVision and Diagnostics also integrate with Business Availability Center, which means you can start with a topology view and drill all the way down to find the status of the most valuable transaction running through you systems.

You can manage what you can’t measure. So what do I do now? If you are properly instrumented the problem will show itself. If you don’t find something you can fix, you can tell the app developers where they need to look to fix the problem.

Related Items:

· End User Monitoring

· Operations Manager

· SiteScope

· SiteScope Administrator Forum

· TransactionVision

· Diagnostics

· Business Availability Center