![]() | |
|
Tivoli performance viewer
Tivoli Performance Viewer (TPV) is included with WebSphere Application Server V7.0 and is used to record and display performance data. Since WebSphere Application Server V6.0, TPV is integrated into the Integrated Solutions Console.
Using Tivoli Performance Viewer, you can perform the following tasks:
Display Performance Monitoring Infrastructure (PMI) data collected from local and remote application servers:
Summary reports show key areas of contention.
Graphical/tabular views of raw PMI data.
Optionally save collected PMI data to logs.
Provide configuration advice through performance advisor section
Tuning advice formulated from gathered PMI and configuration data.
Log performance data
Using TPV you can log real-time performance data and review the data at a later time.
View server performance logs
You can record and view data that has been logged by TPV in the Integrated Solutions Console.
You can use TPV to create summary reports. These reports let you monitor the server's real-time performance and health. TPV enables you to work with the performance modules. With these modules, you can drill down on specific areas of interest, even old logs. Use the log analysis tools to detect trends over time. TPV can also save performance data for later analysis or problem determination.
As the TPV runs inside the Integrated Solutions Console, the performance impact depends on which edition of WebSphere Application Server you run. When running the single server edition, the TPV runs in the same JVM as your application. In Network Deployment, the TPV runs in the JVM of the deployment manager. Certain functions (like the advisor), however, require resources in the node agents or in the application servers.
WebSphere performance advisors
Gathering information made available through the PMI, the WebSphere performance advisors have the ability to make suggestions about the environment. The advisors are able to determine the current configuration for an application server, and trending the PMI data over time, make informed decisions about potential environmental changes that can enhance the performance of the system. Advice is hard coded into the system and is based on IBM best practices for tuning and performance. The advisors do not implement any changes to the environment. Instead, they identify the problem and allow the system administrator to make the decision whether or not to implement. You should perform tests after any change is implemented. There are two types of advisors:
Performance and Diagnostic Advisor
This advisor is configured through the Integrated Solutions Console. It writes to
the SystemOut.log
and to the console while in monitor mode. The interface is
configurable to determine how often data is gathered and advice is written. It
offers advice about the following components:
J2C Connection Manager
Thread pools
LTC Nesting
Serial reuse violation
Plus various different diagnostic advises
Web Container Session Manager
Session size with overflow enabled
Session size with overflow disabled
Persistent session size
Web Container
Bounded thread pool
Unbounded thread pool
Orb Service
Unbounded thread pool
Bounded thread pool
Data Source
Connection pool size
Prepared statement cache size
Java virtual machine (JVM)
Memory leak detection
If you need to gather advice about items outside this list, use the Tivoli Performance Viewer Advisor.
Performance Advisor in Tivoli Performance Viewer
This advisor is slightly different from the Performance and Diagnostic Advisor. The Performance Advisor in Tivoli Performance Viewer is invoked only through the TVP interface of the Integrated Solutions Console. It runs on the application server you are monitoring, but the refresh intervals are based on selecting refresh through the console. Also, the output is routed to the user interface instead of to an application server output log. This advisor also captures data and gives advice about more components. Specifically, this advisor can capture the following types of information:
ORB service thread pools
Web container thread pools
Connection pool size
Persisted session size and time
Prepared statement cache size
Session cache size
Dynamic cache size
JVM heap size
DB2 performance configuration
The Performance Advisor in Tivoli Performance Viewer provides more extensive advice than the Performance and Diagnostic Advisor. Running the Performance Advisor in Tivoli Performance Viewer can require plenty of resources and impact performance. Use it with care in production environments.
WebSphere request metrics
PMI provides information about average system resource usage statistics but does not provide any correlation between the data. Request metrics, in contrast, provide data about each individual transaction and correlate this data.
Request metrics gather information about single transactions within an application. The metric tracks each step of a transaction and determines the process time for each of the major application components. Several components support this transaction metric:
Web server plug-ins
Web container
EJB container
JDBC calls
Web services engine
Default messaging provider
The amount of time that a request spends in each component is measured and aggregated to define the complete execution time for that transaction. Both the individual component times and the overall transaction time can be useful metrics when trying to gauge user experience on a site. The data allows for a hierarchical by response time view for each individual transaction. When debugging resource constraints, these metrics provide critical data at each component. The request metric provides filtering mechanisms to monitor synthetic transactions or to track the performance of a specific transaction. By using test transactions, you can measure performance of the site end-to-end.
From a performance perspective, using transaction request metrics can aid in determining if an application is meeting service level agreements (SLAs) for the client. The metrics can be used to alert the user when an SLA target is not met.
Request metrics help administrators answer the following questions:
What performance area should the user be focused on?
Is there too much time being spent on any given area?
How do I determine if response times for transactions are meeting their goals and do not violate the SLAs?
![]() ![]() ![]() |