BEA has released the JRockit Runtime Analyzer, which is built into the WebLogic JRockit JVM. The tool collects data as your application runs, giving you information that is within the JVM itself. BEA is looking for people to run the analyzer and feed them their output to help tune the JVM even more.
The JRockit Runtime Analyzer capability has been used internally by WebLogic JRockit developers to refine and improve the JVM. However, realizing it's tremendous value, we're now making it available publicly so that all users can also use it to analyze their own Java applications.
In return, we ask for one favor... please help us further improve the WebLogic JRockit JVM. You can do this by sending us an email to "jrockit-improve at bea dot com" with an attachment of the file (in zip format) containing the recording from your own application along with a brief description of that application. This will allow us to analyze the behavior of JRockit under various operating conditions so that we can further improve its runtime behavior for the various types of applications.
However, bear in mind, the JRockit Runtime Analyzer tool is not intended or designed to be a comprehensive application profiling tool. While it may provide some interesting information, it has not been developed with the conveniences and features that may be expected by end users. Your ideas for enhancements are always welcome and will be considered for potential future versions.
Links
Download JRockit Runtime Analyzer Tool
View documentation
-
BEA Releases New Runtime Analyzer Tool (6 messages)
- Posted by: Eric Stahl
- Posted on: July 22 2003 19:27 EDT
Threaded Messages (6)
- BEA Releases New Runtime Analyzer Tool by Vic Cekvenich on July 24 2003 14:20 EDT
- JRockit and M to N thread mapping by Artem Kornilov on July 24 2003 16:19 EDT
-
JRockit and M to N thread mapping by Ricardo Morin on July 24 2003 04:31 EDT
-
JRockit and M to N thread mapping by Artem Kornilov on July 24 2003 05:09 EDT
- Thin threads, Volano mark, and async IO by Sam Pullara on July 24 2003 05:23 EDT
-
JRockit and M to N thread mapping by Artem Kornilov on July 24 2003 05:09 EDT
-
JRockit and M to N thread mapping by Ricardo Morin on July 24 2003 04:31 EDT
- JRockit and M to N thread mapping by Artem Kornilov on July 24 2003 16:19 EDT
- related JSR's by Sean Sullivan on July 25 2003 01:04 EDT
-
BEA Releases New Runtime Analyzer Tool[ Go to top ]
- Posted by: Vic Cekvenich
- Posted on: July 24 2003 14:20 EDT
- in response to Eric Stahl
J:Rockit rocks, it is a great server side VM, I use it and recommend it to my clients. (instead of Sun VM)
I do not have time to create stress tests, when I do, I will try to run some test of basicPortal.com and send in.
Note: I love the j:rockit, but I use it with Resin.
.V -
JRockit and M to N thread mapping[ Go to top ]
- Posted by: Artem Kornilov
- Posted on: July 24 2003 16:19 EDT
- in response to Vic Cekvenich
As far as I know JRockit is using M to N thread mapping in order to improve performance.
So the question is how to force a thread to map 1 to 1? This is required for certain JNI applications for example.
Artem -
JRockit and M to N thread mapping[ Go to top ]
- Posted by: Ricardo Morin
- Posted on: July 24 2003 16:31 EDT
- in response to Artem Kornilov
As far as I know JRockit is using M to N thread mapping in order to improve performance.
>
> So the question is how to force a thread to map 1 to 1? This is required for certain JNI applications for example.
JRockit's default thread model is 1:1 (-Xnativethreads). The M:N model is called Thin Threads (-Xthinthreads), and according to the official documentation it is "experimental."
See:
http://e-docs.bea.com/wljrockit/docs81/userguide/threads.html#1002300
Thanks,
Ricardo -
JRockit and M to N thread mapping[ Go to top ]
- Posted by: Artem Kornilov
- Posted on: July 24 2003 17:09 EDT
- in response to Ricardo Morin
So it works basically all or nothing and there is no facility to do 1 to 1 mapping only for some threads. Well I guess it is a way around the problem.
I brought the whole topic up since the latest VolanoReport (http://www.volano.com/report/) was praising thin threads so much.
Artem -
Thin threads, Volano mark, and async IO[ Go to top ]
- Posted by: Sam Pullara
- Posted on: July 24 2003 17:23 EDT
- in response to Artem Kornilov
Now that async IO is available in the VM there is no reason to run so many threads. The Volano Mark is woefully out of date and not an indicator server JVM performance unless you wrote your server based on the 20 line one in java in a Nutshell. I would look to the newer tests such as SpecJBB for server JVM performance.
Sam -
related JSR's[ Go to top ]
- Posted by: Sean Sullivan
- Posted on: July 25 2003 01:04 EDT
- in response to Eric Stahl
JSR 174
Monitoring and Management Specification for the Java Virtual Machine
http://www.jcp.org/en/jsr/detail?id=174
JSR 163
Java Platform Profiling Architecture
"A mechanism and APIs for extracting time and space profiling information from a running Java virtual machine."
http://www.jcp.org/en/jsr/detail?id=163