New JProbe command-line tools for memory dump analysis and for JProbe analysis engine control

Stuart Hodgins — Thu, 19 Nov 2009 22:01:00 GMT

We are pleased to provide two new command-line utilities that were developed for the next version of JProbe. The first can be used standalone to obtain basic memory usage reports from JVM memory dumps. The second can be used for controlling a JProbe Analysis Engine in a headless environment or for scripting JProbe actions.

JProbe's new JDBC component nodes

Stuart Hodgins — Mon, 05 Oct 2009 18:13:00 GMT

JProbe's new JDBC component nodes help link Java and SQL performance analysis together, to identify the root causes of performance issues in your applications, and to help you resolve those issues.

New in JProbe 8.2.0

Stuart Hodgins — Wed, 30 Sep 2009 15:17:00 GMT

With General Availability of JProbe 8.2, here is a preview of some of the changes and improvements that you will find in this new release:

The JProbe 8.1 release focused on improvements to JProbe Memory analysis. The JProbe 8.2 release focuses on improvements to JProbe Performance analysis.

Users asked us for the ability to track the runtime execution status of threads. JProbe 8.2 introduces a new Threads sub-tab of the Runtime Summary view. During an application analysis session, this tab shows the combined status of all threads running in the JVM, with individual thread details available. This tab replaces the previous Deadlock tab & detection mechanism.

Users asked us for a way to get a broad initial overview of application performance, at less instrumentation cost. We’ve introduced a new “Public methods” instrumentation level that can significantly reduce the method invocations tracked by JProbe, resulting in faster data collection and less data complexity. As always, setting appropriate data collection filters remains key to getting the best data at reasonable collection overhead.

Users have often asked us how JProbe can help to visualize and diagnose database connection and usage issues within their Java applications. JProbe 8.2 introduces a new automated JDBC data collection option. Using technology patterned on the successful PerformaSure approach, JProbe will instrument the JDBC “component boundary”. Combined with standard application filters, this feature shows where JDBC calls led to database activity, and shows those calls in the calling context, that is, shows the sequence of method invocations that caused the db access to occur.

Users asked us for a way to immediately focus on the likely candidates for performance improvement. The new Summary view provides a list of top methods to investigate, and uses these methods to present a smaller and more focused Call Graph for detailed analysis. If you want to start your analysis from a particular known method, you can easily do that too. Simply enter the method name and JProbe will prompt you to select the correct one from a list of matches.

Related to this change, a new “add more nodes” feature on the Methods view’s Call Graph tab replaces the former “Show top N” capability, and intelligently expands the Call Graph to show the execution context of the focus node.

Users have often asked us for elapsed and CPU times together on the same view. JProbe 8.2 provides this option whenever CPU data has been collected. You can display the detailed data side-by-side on the Methods view, or easily focus on either type of data with a click.

An updated Methods view (formerly the Call Graph view) provides an additional level of detail in the new two-level Methods list. The top tier corresponds to the aggregate Call Graph view of a method, while the second tier of data directly corresponds to the set of Call Tree invocations along different call paths.

Users asked us how to see which methods could have contributed to a Grouped node when using the default “Grouped” encapsulation level. A new “Show Method Names” context menu entry provides this data.

These new features and other improvements make us excited about the JProbe 8.2 release. We look forward to your feedback on them and to your suggestions for future improvements!

JProbe Blog

New JProbe command-line tools for memory dump analysis and for JProbe analysis engine control

JProbe's new JDBC component nodes

New in JProbe 8.2.0