Troubleshoot performance

Troubleshoot performance

This table describes some common performance problems and the action that you can take to resolve them.

Problem User action

ORB: Response time and throughput indicate that enterprise bean requests with shorter processing times are being denied adequate access to threads in ORB thread pool. Set the Logical Pool Distribution (LPD) mechanism for the ORB service. See Object Request Broker service custom properties for more information.

Under a load, client requests do not arrive at the Web server because they time out or are rejected. For IBM HTTP Server for iSeries (powered by Apache), set the ListenBackLog parameter.

The Percent Maxed metric from Tivoli Performance Viewer indicates that the Web container thread pool is too large or too small. For information about the Percent Maxed metric, see Thread pool on PMI data counters. Set the Thread pool Maximum size for the Web container.

Netstat shows too many TIME_WAIT state sockets for the application server's internal HTTP port (9080 is the default port). Set the HTTP transports MaxKeepAliveConnections and HTTP transports MaxKeepAliveRequests properties. See Set custom properties for an HTTP transport for more information.

Disk arm utilization is too high, due to excessive paging. Verify that there is enough memory in the pool in which WebSphere Application Server is running, and that the Java virtual machine heap is not leaking. For more information, see Java virtual machine tuning parameters and Java memory tuning tips.

The Percent Used metric for a data source connection pool from Tivoli Performance Viewer indicates the pool size is too large. For information about the Percent Used metric, see JDBC connection pools on PMI data counters. Set the Maximum connection pool and Minimum connection pool properties. See Connection pool settings for more information.

The Prepared Statement Discards metric from Tivoli Performance Viewer indicates that the data source statement cache size is too small. For information about the Prepared Statement Discards metric, see JDBC connection pools on PMI data counters. Set the Statement cache size property. See Data source settings for more information.

The Percent Maxed metric from Tivoli Performance Viewer indicates that the Object Request Broker thread pool is too small. For information about the Percent Maxed metric, see Thread pool on PMI data counters. Configure EJB method Invocation Queuing.

The server is unresponsive and paging occurs. Check for memory leaks and verify that enough memory is available. See Java memory tuning tips for more information.

Throughput, response time and scalability are undesirable. Use the dynamic cache service.

Problem	User action
ORB: Response time and throughput indicate that enterprise bean requests with shorter processing times are being denied adequate access to threads in ORB thread pool.	Set the Logical Pool Distribution (LPD) mechanism for the ORB service. See Object Request Broker service custom properties for more information.
Under a load, client requests do not arrive at the Web server because they time out or are rejected.	For IBM HTTP Server for iSeries (powered by Apache), set the ListenBackLog parameter.
The Percent Maxed metric from Tivoli Performance Viewer indicates that the Web container thread pool is too large or too small. For information about the Percent Maxed metric, see Thread pool on PMI data counters.	Set the Thread pool Maximum size for the Web container.
Netstat shows too many TIME_WAIT state sockets for the application server's internal HTTP port (9080 is the default port).	Set the HTTP transports MaxKeepAliveConnections and HTTP transports MaxKeepAliveRequests properties. See Set custom properties for an HTTP transport for more information.
Disk arm utilization is too high, due to excessive paging.	Verify that there is enough memory in the pool in which WebSphere Application Server is running, and that the Java virtual machine heap is not leaking. For more information, see Java virtual machine tuning parameters and Java memory tuning tips.
The Percent Used metric for a data source connection pool from Tivoli Performance Viewer indicates the pool size is too large. For information about the Percent Used metric, see JDBC connection pools on PMI data counters.	Set the Maximum connection pool and Minimum connection pool properties. See Connection pool settings for more information.
The Prepared Statement Discards metric from Tivoli Performance Viewer indicates that the data source statement cache size is too small. For information about the Prepared Statement Discards metric, see JDBC connection pools on PMI data counters.	Set the Statement cache size property. See Data source settings for more information.
The Percent Maxed metric from Tivoli Performance Viewer indicates that the Object Request Broker thread pool is too small. For information about the Percent Maxed metric, see Thread pool on PMI data counters.	Configure EJB method Invocation Queuing.
The server is unresponsive and paging occurs.	Check for memory leaks and verify that enough memory is available. See Java memory tuning tips for more information.
Throughput, response time and scalability are undesirable.	Use the dynamic cache service.