Monitor overall system health
Overview
Systems to monitor include...
- Web servers
- appservers
- databases
- back-end systems
- other critical systems
If any system has a problem, it might cause the message...
servlet is slow...to appear.
WAS (WAS) can generate PMI data, which provides average statistics on...
- WAS resources
- application resources
- system metrics
Monitor Task List
Start monitoring as soon as you have applications to monitor. Establish a baseline.
- Enable PMI to begin data collection.
You can enable with scripting.
- Use Tivoli Performance Viewer or third-party solutions to monitor performance.
- Extend PMI or develop your own monitoring applications
What to monitor
Monitor the following statistics at a minimum:
Metric Description Average response time For example,...
- servlet response time
- enterprise beans response time
...where response time statistics indicate how much time is spent in various WAS systems.
Number of requests (transactions) How much traffic is processed by WAS. Used for capacity management. Live HTTP sessions Concurrent usage of the site. The more concurrent live sessions, the more memory is required. As number increases, try adjusting session time-out or the available JVM heap.
Web server thread pools
Web thread pools
EJB thread pools
ORB thread pools
Database connection poolsSetting too large impacts the amount of memory that required
Setting too small might cause bottlenecks if downstream resources can handle workload increases.
Java virtual memory Use this metric to help set the optimal heap size and identify potential memory leaks. Monitor frequency of garbage collection. CPU
I/O
System pagingSystem resources that affect workload capacity.
See also
Performance Monitoring Infrastructure
Custom PMI API
Enable PMI data collection
Develop your own monitoring applications
Monitor performance with Tivoli Performance Viewer
Third-party performance monitoring and management solutions