1	Profiling Java applications using Eclipse* Test and Performance Tools Platform
2	Instructors Mikhail Voronin Intel Corporation mikhail.voronin@intel.com Alexander Alexeev Intel Corporation TPTP Committer alexander.n.alexeev@intel.com
3	Agenda Unit 1 – Eclipse* TPTP Overview Unit 2 – Using the TPTP Java Profiler Profiler Architecture Profiling and Logging Perspective Launch and Attach Profiling options and views Demo Profiler Preferences Conclusion, Opens and Limitations
4	Participants Tell me a little about yourselves… From where Company or academic institution Users or Adopters Eclipse* and TPTP experience Java* experience Have you used a Java* profiler before? Tutorial expectations
5	"Unit 1 – Eclipse" Unit 1 – Eclipse TPTP Overview Unit 2 – Using the TPTP Java Profiler Profiler Architecture Profiling and Logging Perspective Launch and Attach Profiling options and views Demo Integration with BIRT Profiler Preferences Conclusion, Opens and Limitations
6
7	Eclipse* TPTP Overview Composed of 4 sub projects: Platform Test Trace and Profiling Monitoring Principles: Extension of the Eclipse* Value Proposition Vendor Ecosystem Vendor Neutrality Standards-Based Innovation Agile Development Inclusiveness & Diversity TPTP is highly extensible.
8
9	Eclipse* TPTP Profiling Tool Overview Broadly useful for performance analysis and for gaining a deeper understanding of a Java* program Consists of the Profiling and Logging Perspective and a number of graphical and tabular views Visualize program execution and threading behavior; Pinpoints operations that taking most resources; Explore patterns of program behavior Enables you to test your application's performance early in the programming development cycle for improvements Option to run application with agent at near full speed (enabled mode) and attach to it to gathering details at certain phase. Assorted filtering functionality which can help to localize problem and reduce overhead for large long run applications
10	Eclipse* TPTP Agent Controller Overview A daemon process that resides on each deployment host and enables client applications to launch host processes and interact with agents that coexist within host processes Contains a server that has the capability to launch and manage local or remote applications from a local TPTP workbench The Java* profiler is a managed agent that can be used to profile local or remote Java* applications from a local TPTP workbench
11	JVMTI-based Java* Profiler vs. JVMPI-based Java* 5.0 has introduced new standards for profiling Prior to Java* 5.0 (Java* 1.4-) the standards and the interfaces to support profiling (JVMPI) were experimental Java* J2SE 6 eliminated JVMPI interface The new standard (JVMTI) is an innovative solution to profiling and enables you control precisely which parts of an application are profiled Eclipse* TPTP has embraced this innovation and the new Java* profiler is based on this new standard In Eclipse* TPTP old Java* profiler (JVMPI) still exists Only for backward compatibility with Java* 1.4 Because of JVMPI/JVMTI interface differences there are some differences in behavior between old and new profiler Old Java* profiler will eventually be removed from Eclipse* TPTP
12	"Unit 1 – Eclipse" Unit 1 – Eclipse TPTP Overview Unit 2 – Using the TPTP Java* Profiler Profiler Architecture Profiling and Logging Perspective Launch and Attach Profiling options and views Demo Integration with BIRT Profiler Preferences Conclusion, Opens and Limitations
13	JVMTI-based Java* Profiler A set of libraries that attach to a JVM for recording Java* application's behavior An extensible framework, consisting of a core runtime component (Martini), an agent managed by the Agent Controller (JPIAgent), and a set of data collection libraries built on top of the Martini runtime. Can be used to identify performance details such as classes or methods responsible for execution bottlenecks, analyze application heap to find memory leaks and visualize threading behavior. Output in the form of XML fragments (XML4Profiling) Can be launched from the Eclipse* IDE or as a standalone program using JVM command-line options Applications under test can reside in Eclipse* workspace, binaries on file system, or hosted in a J2EE application server.
14	JVMTI-based Java* Profiler Architecture
15	Profiling and Logging Perspective Profiling and Logging perspective provides resources for starting a profiling session as well as obtaining comprehensive information on the performance of the monitored application The profiling tool provides information pertaining to Execution analysis Method execution performance Object allocations Thread interactions
16	Profiling and Logging Perspective Combinations of views and editors that are best suited to perform application profiling Profiling Monitor view Administrate profiling resources and manage activities Profiling views Visualize and analyze profiling data Profiling actions Control profiling resources Actions are resource sensitive Choice of action depending on type and status of the object in selection Attach and Detach the agent from process Start and Stop monitoring on an agent Terminate a process
17
18
19	Launch and Attach Profiling session is started by launching an application or by attaching to a running application Launch : start an application with a profiling agent Attach : attach to an application which is already started and invoked with a profiling agent How? Launch configuration is where you start the session.
20	Launch and Attach - Configuration A place to create, run and manage profiling sessions. Configure the details of a profiling session Target host Application to profile Scope of the profile Profile data destination Select launch configuration type according to the location of the target Java* application: Within Workbench > “Java Application” launch configuration Outside Workbench > “External Java Application” launch configuration
21	Launch and Attach – Configuration Tabs Collect and display detail information about a configuration Tabs for profiling: Monitor tab : defines detail of data collection on a profiling session Destination tab : defines the destination of the profiling data Host tab : for external application configurations only, defines the location of process to be launched or attached Agents tab : for attach configurations only, lists agents available for attach
22	Launch and Attach – Monitor Tab
23	Launch and Attach – Profiling Options
24	Profiling Options – Filter Set Limit the scope of a profiling session Ensure that only relevant details are collected Especially useful when speed and efficiency is critical Excluded classes and methods are not instrumented and execute in full speed Only the first applicable filter is applied. When you are specifying filters, ensure that you declare the most specific filter criteria first
25	Launch and Attach – Execution Analysis Options
26	Execution Time Analysis Options Specifies the type of information to collect during execution time analysis Use the “Show execution statistics” option to identify the most time-consuming methods Low-overhead. Can be used without extensive filtering Use the “Show execution flow graphical details” option to identify the relationships between executing methods (call-tree). High-overhead. Filtering is recommended.
27	Launch and Attach – Memory Analysis Options
28	Memory Analysis Options Specifies the type of information to collect during memory analysis Use the “Track object allocation sites” option to identify the source line where each object is allocated May slightly increase analysis overhead
29	Launch and Attach – Destination Tab Specify the destination of the profile data Workbench : Visualize and analyze in profiling views (default) File : export to XML file, available for import. Import profiling file File > Import ... > Profiling file
30	Profiling an External Application – Host Tab
31	Attaching to a Running Application – Monitor Tab
32	Profiling Views A collection of views to visualize and analyze profiling data Memory Statistic view (tabular) Execution Statistic view (tabular) Coverage Statistic view (tabular) Method Invocation Detail view (tabular) Execution Flow view and table (graphical + tabular) Method Invocation view and table (graphical + tabular) Execution Call Tree (graphical + tabular) UML2 Trace Interactions view (graphical)
33	Recommendations for effective profiling Using filters to collect only needed data Pause/resume to select activities to profile Attach/Detach to eliminate overhead when we don’t profile
34	A few words on filtering in TPTP Collection Time Filtering Pro: Reduce data volume and instrumentation overhead (critical) Con: Data for excluded classes/method can never be seen Model Import Filtering Pro: Reduce workbench memory footprint (critical on 32bit OSes) Con: Can only see data that is in the model View Level Filtering Pro: Remove items from being visualized in the workbench Con: Does not reduce data volume or model footprint
35	Access to Attach/Detach and Pause/Resume
36	Java* Profiling - Demo Profiling a local Java* application Execution Time Analysis Memory Analysis Thread Analysis Profiling an External Java* Application Profiling an Eclipse* Plugin (RCP Application) Profiling a J2EE (Server) Application
37	Java* Profiling - Preference Dialog to set user preference Window > Preferences... Set default profiling values Add host Configure view preferences
38	Java* Profiling - Attach Start application under profile in standalone mode Invoke the Java* Profiling Agent, a library that attaches to a Java* virtual machine (JVM) to capture and record the behavior of the Java* application Use the -agentlib JVM option to invoke the Java* Profiling Agent: -agentlib:JPIBootLoader=JPIAgent:server=enabled;{profiler} Communication with the agent invoked is done using a client workbench by means of the Agent Controller on the host machine. Attach to the agent : Profile Configuration... > Attach to Agent
39	"Unit 1 – Eclipse" Unit 1 – Eclipse TPTP Overview Unit 2 – Using the TPTP Java* Profiler Profiler Architecture Profiling and Logging Perspective Launch and Attach Profiling options and views Demo Integration with BIRT Profiler Preferences Conclusion, Opens and Limitations
40	TPTP Profiler Future Updates for Ganymede Improve overall usability via assorted fixes Using feedback from users Reduce usability diffs between JVMPI and JVMTI… e.g., SSL for secure communications e.g., Reduce overuse of dynamic attach/detach Enhancements e.g., Better multithread analysis Improved visualization Contended lock analysis Track more thread states (e.g., join points) e.g., Efficient binary transfer format e.g., Better Java* 6 support Long-term JVM Heap analysis and trends Object reference analysis Graphic representation of profiling data and JVM telemetry
41	Resources
42	"Q & A" Q & A
43	TPTP @ EclipseCon2008 Short Tutorials (2 hours) Profiling Java* applications using Eclipse* TPTP Alexander Alexeev, Mikhail Voronin, Monday, 08:00, Great America Meeting Room 1 Long Talk (50 mins) A Harmonious Combo : Experiences Profiling Apache* Harmony with Eclipse* TPTP Chris Elford, Sergey Kuksenko, Thursday, 14:30, Grand Ballroom C Functional Testing with the Eclipse* TPTP Project Eric Long, Tuesday, 14:30, Room 206 Extending TPTP for database unit testing Wei Liu, Der Ping Chou, Hong-Lee Yu, Don Clare, Thursday, 13:30, Theatre Short Talk (10 mins) Extending TPTP for TTCN-3. Paul Slauenwhite, Wednesday, 15:30, Room 209/210
44
45	"Legal Notices:" Legal Notices: Other company, product, or service names may be trademarks or service marks of others.