Production JVM Monitoring

Production monitoring can be very 'expensive' in terms of performance, but staging environments don't always reproduce issues. And, when a production application runs into problems, you can't turn on lab visibility tools - adding instrumentation will only make the problem worse. This lack of visibility delays problem resolution. It's important to choose the right tool for the issue at hand. The IT Toolbelt shows some common tools, most of which have significant performance impact on running systems. Some low-impact choices include asynchronous logging, jconsole and Azul Zing Vision (part of the Zing Platform).

IT Toolbelt:  Some Monitoring Tools        

Zing Vision Monitoring: Finding Production Problems

To provide production JVM monitoring, the Zing Platform includes Zing Vision for non-intrusive, zero-overhead visibility. Below are some of the JVM problems Zing Vision can help diagnose:

• Java memory leaks
  • Symptoms: Application consumes all available heap memory, live heap trend is a ramping sawtooth
• Locks
  • Symptoms: Adding users/threads/CPUs causes the app to slow down; high lock acquire times and contention; race conditions, deadlock, I/O under lock
• Endless Exceptions
  • Symptom: Application spends time in j.i.t.fillInStackTrace()

More on this Topic:

• Presentation on Resolving Top Java Production Issues »
• Gil Tene's Webinar on Zing Vision »
• The Zing Vision production time diagnostic and tuning tool »

 Back to the Zing How it Works Page »