In this release of WebSphere Application Server Advanced Edition, Resource Analyzer is a technology preview. Consult the README file for details on known problems and limitations. This file introduces the Resource Analyzer and contains the following topics:
The Resource Analyzer is a stand-alone performance monitor for WebSphere Application Server Version 3.5 Advanced Edition. The Analyzer retrieves performance data by periodically polling the administrative server. Data is collected continuously and retrieved as needed from within the Analyzer. The level of data to collect is specified by using the WebSphere Advanced Administrative Console. Then, the Analyzer's graphical interface is used to retrieve and view the data in a table or chart, or to store the data in a log file.
Analyzer provides a wide range of performance data for two kinds of resources: WebSphere resources (for example, enterprise beans and servlets) and run time resources (for example, Java Virtual Machine (JVM) memory, application server thread pools, and database connection pools). Performance data includes statistical data (such as the response time for each method invocation of an enterprise bean) and load data (such as the average size of a database connection pool during a specified time interval). This data is reported for individual resources and aggregated for multiple resources. The Analyzer collects and reports performance data for the following resources:
Depending on which aspects of performance are being measured, the Analyzer can be used to do the following:
With this data, the Analyzer can be used to do the following:
Data collection and reporting incurs a cost--it can affect the performance of your distributed applications. Although performance data is automatically collected at all times, it is not reported by default. Instead, you must explicitly enable data reporting for those aspects of your system to be monitored. Even when data reporting is enabled, the types of data displayed depend on user-defined instrumentation levels. These levels allow you to choose among high-impact and low-impact data, and thus control the overall performance impact of the data-collection process itself. See Setting levels for data collection.
Resource Analyzer organizes performance data in a hierarchy of groups. A group is a set of statistics, or counters, associated with a particular resource. Counters measure some aspect of a running system, for example, the number of active enterprise beans, the time spent in responding to a servlet request, or the number of kilobytes of memory. The database connection pool group is a set of related counters for measuring connection pool activity. Counters are said to be members of the group.
Groups can have subgroups. Counters can belong to any number of groups, or to no group at all. For example, the counter methodRt (method response time) belongs to the enterprise bean group and to the container group. (The counter represents a count for the individual bean and an aggregate count for all beans in a container, respectively.) Groups can belong to other groups. The diagram in Figure 1 shows a hierarchy of groups.
Figure 1. Example hierarchy of performance groups
![]() |
The enterprise beans performance category is a root group. Subgroups (descendants) of this group are container instances. Each container instance represents the set of counters for that container. Counters for containers include the number of method calls and their average response time for all enterprise beans in the container. Enterprise beans residing in a container belong to a subgroup whose name is the remote class name of the enterprise bean. In this example, the bean2 and bean3 groups belong to both the container1 and container2 groups.
When performance data is displayed for a group, the information is organized into families. Families are simply a view of all similar objects--family tables show multiple resources of the same type and their counters. A family table for entity beans consists of the names of all entity beans in a container and the values for their corresponding counters. A family table for session beans consists of all session beans and their corresponding counters.