Network Deployment (Distributed operating systems), v8.0 > Tune performance > Tune the application serving environment
Tune the application server using pre-defined tuning templates
Overview
We can use the python-based tuning script, applyPerfTuningTemplate.py, along with one of its template files, to apply pre-defined performance tuning templates to the application server or cluster. The script, and these property-based template files are located in...
-
WAS_HOME/scriptLibraries/perfTuning/V70
The configuration settings applied by this script and the associated tuning templates should be viewed as potential performance tuning options for you to explore or use as a starting point for additional tuning. The configuration settings that each of the pre-defined templates applies are geared towards optimizing common application server environments or scenarios. Typically, these settings improve performance for many applications.
Because optimizing for performance often involves trade-offs with features, capabilities, or functional behavior, some of these settings might impact application correctness, while other settings might be inappropriate for the environment. Please review the documentation below and consider the impact of these settings to the application inventory and infrastructure.
As with any performance tuning exercise, the settings configured by the predefined templates should be evaluated in a controlled preproduction test environment. You can then create a customized template to refine the tuning settings to meet the specific needs of the applications and production environment. bprac
Typically, when you run the applyPerfTuningTemplate.py script, you will specify either the production.props template file or the development.props template file to apply against the target server or cluster.
- If you specify the production.props template file when you run the applyPerfTuningTemplate.py script, the script applies configuration settings that are appropriate for a production environment where application changes are rare and optimal runtime performance is important.
- If you specify the development.props template file when you run the applyPerfTuningTemplate.py script, the script applies configuration settings that are appropriate for a development environment where frequent application updates are performed and system resources are at a minimum.
In addition to these two common templates, a third template file, default.props, is provided to enable you to revert the server configuration settings back to the out-of-the-box defaults settings.
We can also create your own custom tuning template.
To create a custom tuning template, copy one of the existing templates, modify the configuration settings to better fit the needs of the applications and environment, and then use the applyPerfTuningTemplate.py script to apply these customized settings. The script and properties files leverage the property file configuration management features that wsadmin provides, and can easily be augmented to tune additional server components. See the topic Using properties files to manage system configuration for more information.
Review the following table to see the configuration changes that occur based on the template file that you specify when you run the applyPerfTuningTemplate.py script. A blank cell in this table indicates that the listed parameter is not configured, or is configured back to the default settings for the server defaults.
| Parameter | Server default (default.props template file) | Production environment (production.props template file) | Development environment (development.props template file) |
|---|---|---|---|
| JVM Heap Size (MB) | 50 min / 256 max | 512 min / 512 max | 256 min / 512 max |
| Verbose GC | disabled | enabled | enabled |
| JVM Arguments*** | -Dcom.ibm.xml.xlxp.jaxb.opti.level=3 | -Dcom.ibm.xml.xlxp.jaxb.opti.level=3 | -Dcom.ibm.xml.xlxp.jaxb.opti.level=3 |
| HTTP (9080) and HTTPS (9443) Channel maxKeepAliveRequests | 100 | 10000 | 10000 |
| TCP Channel maxOpenConnections | 20000 | 500 | 500 |
| TCP Channel listenBacklog | 511 | 128 | 128 |
| Development Mode | disabled | enabled | |
| Server Component Provisioning | disabled | enabled | enabled |
| PMI Statistic Set | basic | none | none |
| Authentication Cache Timeout | 10 minutes | 60 minutes | 60 minutes |
| Data Source Connection Pool Size* | 1 min / 10 max | 10 min / 50 max | |
| Data Source Prepared Statement Cache Size* | 10 | 50 | |
| ORB Pass-by-Reference** | disabled | enabled | enabled |
| Web Server Plug-in ServerIOTimeout | 900 | 900 | 900 |
| Thread Pools (Web Container, ORB, Default) | 50 min / 50 max, 10 min / 50 max, 20 min / 20 max | 5 min / 10 max |
* Indicates items that are tuned only if they exist in the configuration. For example, a data source connection pool typically does not exist until an application is installed on the application server. If these items are created after your run the script, they are given the standard server default values unless you specify other settings.
** Enabling ORB Pass-By-Reference can cause incorrect application behavior in some cases, because the Java EE standard assumes pass-by-value semantics. However, enabling this option can improve performance up to 50% or more if the EJB client and server are installed in the same instance, and the application is written to take advantage of these feature. The topic ORB service settings can help you determine if this setting is appropriate for the environment.
*** This setting might cause issues when web services are used in certain scenarios. Therefore, if you are running web services, and are not experiencing throughput optimization issues, you can remove this parameter from the script, or set the opti level to 0.
Following are a few subtle platform-specific tuning differences:
- -XX:-UseAdaptiveSizePolicy
- -XX:+UseParallelGC
- -XX:+AggressiveOpts
- -XX:+UnlockDiagnosticVMOptions -server
- -Dcom.ibm.xml.xlxp.jaxb.opti.level=3
- -XX:+AggressiveOpts
- -XX:+ForceMmapReserved
- -XX:SurvivorRatio=16
- -Xoptgc
- -XX:+UseParallelGC
- -Djava.nio.channels.spi.SelectorProvider=sun.nio.ch.DevPollSelectorProvider
- -XX:-ExtraPollBeforeRead -XX:+UseSpinning
- -Dcom.ibm.xml.xlxp.jaxb.opti.level=3
Procedure
- Start wsadmin.sh if it is not already running, and then complete one of the following actions to tune an application server or all of the application servers in a cluster.
- Run the applyPerfTuningTemplate.py script to tune a specific server or cluster of servers running in a production
environment.
wsadmin -f applyPerfTuningTemplate.py [-nodeName node_name -serverName server_name
[clusterName cluster_name] -templateFile production.props
- Run the applyPerfTuningTemplate.py script to tune a specific server or cluster of servers running in a development
environment.
wsadmin -f applyPerfTuningTemplate.py [-nodeName node_name -serverName server_name
[clusterName cluster_name] -templateFile development.props
- Run the applyPerfTuningTemplate.py script to change the settings for a server or a cluster back to the standard
out-of-the-box default configuration settings.
wsadmin -f applyPerfTuningTemplate.py [-nodeName node_name -serverName server_name
[clusterName cluster_name] -templateFile default.props
What to do next
Conduct a performance evaluation, and tuning exercise
to determine if you should further fine tune the server for your specific
applications.
Tune application servers
Tune the IBM virtual machine for Java
Enable PMI data collection
Use properties files to manage system configuration
Enable PMI data collection
Tune the application serving environment
Related
Java virtual machine custom properties
HTTP transport custom properties
Application server settings
Authentication cache settings
Connection pool settings
WAS data source properties
Thread pool settings
ORB service settings