The purpose of this Wiki page is to summarize the resources and tips for profile data analyzer which is a standalone tool to systematically analyze issues from profiling data files (profiling files from SAP JVM Profiler or ABAP Runtime Analysis) and create analysis reports. These analysis reports can be used by application developers, system administrators, and support engineers to easily identify the most expensive parts of a program and optimize them accordingly.
Here are the typical scenarios for the profile data analyzer.
- Visualize ABAP Runtime Analysis or SAP JVM Profiler profiling files as interactive and intuitive Flame Graphs,
- Graphically compare performance traces in order to e.g. to understand a performance degradation after implementing a Support Package,
- Convert ABAP Runtime Analysis files into the SAP JVM Profiler format to make use of advanced profiling features of SAP JVM Profiler
The following SCN blog contains a getting started guide for profile data analyzer. When you download the package, you will also find a PDF guide named "Profile_Data_Analyzer.pdf" together with the executable java file.
As shown in the above blog posts, the performance flame graph is one of the most interesting parts in the analysis report of profile data analyzer. Here is an example flame graph based on a SM21 ABAP trace and the tips about how to read the performance flame graph.
- Every column is a call stack. The main program is at the bottom. The lower method calls the upper method.
- The width in the graph is proportional to the actual time used by that call stack.
- We could zoom in or zoom out in the graphic view by clicking a method on the call stack.
- Click the "Search" button to search with regular expressions. The search results will be highlighted in purple.
- Move the mouse to a method, the tooltip will show more detailed time information.
- Special method types have dedicated colors. For example, DB methods are blue.
- The call count of a call stack is shown like "Calls: XXX" on top of the call stack.
Here are the KBAs for the profile data analyzer, including how to collect the trace and how to analyze it with profile data analyzer.
- KBA 2879724 - How to analyze performance traces with the profile data analyzer
- KBA 2881237 - How to collect performance trace for the profile data analyzer
- KBA 2891307 - How to convert ABAP trace by profile data analyzer and analyze it with SAP JVM Profiler
Here are the documents for SE30, SAT and ABAP performance trace
- Executing Measurements in SE30/SAT: Online help link
- Starting the ABAP Profiler on ABAP in Eclipse: Online help link
- Aggregation by call stack which could reduce the trace size and only available on ABAP in Eclipse: Online help link
Here are the documents for the SAP JVM Profiler. To lever the full potential of the SAP JVM Profiler, we suggest going through the following SAP JVM Profiler documentation.
Support & Feedback
Your feedback is need and appreciated, and we are looking to your feedback. But the support channel of the profile data analyzer is not the OSS incident in the SV-PERF component. If you have any feedback or any issue while using the profile data analyzer, please feel free to send an email to "qiansheng.wang AT sap.com" or post a comment in the following SCN blog post. We will follow up with you there.