Give your system a well-check to keep it in top shape. Focus your attention on these 13 metrics.
-
Web User Count per Instance adds the number of unique sessions originating from available web server instances. The recommended threshold is 40–50. If your count is above 50, you may need to add a server.
Application Management module > Historical Utilization > User Activity -
Excessive numbers of Open Applications Per Session use up resources on the HTML server and can affect overall user performance. Our CNCs recommend no more than 10.
Application Management module > Historical Utilization > User Activity -
Java Heap Used per Instance records the percentage of memory in use by the Java Virtual Machine (JVM) on the web server. The recommended threshold is 75%. If greater than 75%, you may need to increase your heap size or add HTML instances to handle the load.
Application Management module > Historical Utilization > HTML Server -
Server Memory Used per Instance charts the total memory used by the Enterprise Server instance. Watch this number to catch when you need to increase memory.
Application Management module > Historical Utilization > Enterprise Server -
Server CPU Utilized per Instance shows the total percentage of server CPU used by the Enterprise Server instance. Several issues can cause the CPU percentage to rise, but most often, UBEs are to blame. If too many UBEs are submitting to a single server, you may need to move the UBE load.
Application Management module > Historical Utilization > Enterprise Server -
A three or more in the Manual Open JDB Transactions chart means database activity is increasing. Learn how to lower the number here.
Application Management module > Historical Utilization > System Activity -
Zombie Process per Instance should always be zero. If not, gather the log file to see what sent a process into a zombie state. Perhaps the process was killed, or there may be a code issue.
Application Management module > Historical Utilization > System Activity -
Outstanding Requests per Enterprise Server are waiting to be processed by the kernel. A high number may mean there is an excessive load on the system or an issue with the kernel code. Consider increasing the number of kernels per kernel type. For example, you might increase the number of call object kernels that are starting.
Application Management module > Historical Utilization > System Activity -
Average Call Object Time per Instance shows the average amount of time it takes to perform business function calls per web server. Time above the average may indicate problems with a specific BSFN or overall performance issues with the Enterprise Server.
Application Management module > Historical Utilization > Call Object Kernels -
A business function can incur Call Object Errors. If found, gather the log files for the specific function involved to troubleshoot the issue.
Application Management module > Historical Utilization > Call Object Kernels -
Keep the number of Users per Call Object Kernel between five and eight. If you have more than eight users per call object, consider increasing the number of call object kernels on your server.
Application Management module > Historical Utilization > Call Object Kernels -
UBE Analysis—Top 10 Jobs ranks the longest-running JDE report engines. Perhaps a SQL select statement is causing high runtime. In that case, indexing would help resolve the issue. However, you can visit My Oracle Support for more insight on gathering debug logs and using JDE Performance Workbench to analyze the UBE further.
Batch Jobs module -
Scan the List of Page Views table for slow applications. If you do find any, you can learn how to troubleshoot them here.
Analytics module > Page Views
Remember to subscribe to personal alerts for instant notification on these and other critical metrics.