You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@harmony.apache.org by Stefano Mazzocchi <st...@apache.org> on 2005/10/17 23:37:35 UTC
[Fwd: TALK:Wednesday 10-26-05 End-to-End Performance Optimization
of Java Server]
for those in cambridge, ma or surroundings.
--
Stefano.
Re: [Fwd: TALK:Wednesday 10-26-05 End-to-End Performance
Optimization
of Java Server]
Posted by ac...@apache.org.
I'm asking the wife for permission to make a day trip...
Geir Magnusson Jr. wrote:
> Hey, I can be there, it seems. Who else? We can get together for
> lunch beforehand...
>
> geir
>
> On Oct 17, 2005, at 10:37 PM, Stefano Mazzocchi wrote:
>
>> for those in cambridge, ma or surroundings.
>>
>> --
>> Stefano.
>>
>>
>> From: CSAIL Event Calendar <ev...@csail.mit.edu>
>> Date: October 17, 2005 5:23:50 PM BDT
>> To: seminars@csail.mit.edu
>> Subject: TALK:Wednesday 10-26-05 End-to-End Performance Optimization
>> of Java Server
>>
>>
>>
>> End-to-End Performance Optimization of Java Server Workloads
>> Speaker: Jong-Deok Choi
>> Speaker Affiliation: IBM T. J. Watson Research Center
>> Host: Martin Rinard
>> Host Affiliation: CSAIL
>>
>> Date: 10-26-2005
>> Time: 2:00 PM - 3:00 PM
>> Refreshments: 1:45 PM
>> Location: 32-D463 Star
>>
>> The size and complexity of large-scale commercial middleware systems,
>> such as J2EE, and their workloads makes it difficult to understand
>> and optimize the performance of such systems. A J2EE middleware with
>> its application may have tens of thousands of methods, hundreds of
>> concurrently executing threads, and runtime stacks that are hundreds
>> of methods deep. While acting like an operating system for its
>> applications and providing various runtime services, J2EE runs on a
>> Java Virtual Machine (JVM) as an application mostly written in Java.
>> The JVM itself provides its own various runtime services, such as
>> synchronization and garbage collection, and in turn relies on the
>> underlying operating system for OS services such as threading and
>> memory management. This complicated interaction among the various
>> layers of the software (and hardware) stack is a major source of the
>> challenge to understanding and optimizing the performance of such
>> large-scale middleware systems and their applicat
>> ions.
>>
>> In this talk, I will first present a brief description of the whole-
>> stack, end-to-end analysis and optimization system we have developed
>> at IBM T. J. Watson Research Center for Websphere Application Server
>> (WAS), which is IBM's implementation of the J2EE. After that, I will
>> present runtime data and their analysis results from the various
>> layers of the SW and HW stack running WAS and its key benchmark
>> programs, and compare them with those of SPECjbb, a Java benchmark
>> program widely used by Java performance researchers. I will also
>> present a few of the performance bottlenecks we have found, and their
>> optimization.
>>
>> Bio:
>> Jong-Deok Choi is a Research Staff Member (RSM) and Manager of the
>> High Performance Programming group at the IBM T. J. Watson Research
>> Center in Yorktown Heights, New York. He is a project co-leader of
>> the End-to-End Server Optimization project, a cross-organizational
>> effort in IBM for improving the performance of commercial eServer
>> applications on IBM's server platforms. Since joining Watson Research
>> in 1989, he has been involved in various research and development
>> projects - in the areas of program analysis, optimization, and
>> debugging - such as PTRAN, TPO, Jalapeno, DejaVu, JikesRVM, and
>> End-to-End Optimization. His research interests include static and
>> dynamic analysis and optimization of commercial and scientific
>> applications, middleware analysis and optimization, high performance
>> computer architecture, and debugging multithreaded and distributed
>> applications.
>>
>> He received B.S. in Electronic Engineering from SNU, Korea in 1979,
>> M.S. in Electrical Engineering from KAIST, Korea in 1981, and M.S.
>> and Ph.D. both in Computer Sciences from Univ. of Wisconsin-Madison
>> in 1985 and 1989, respectively.
>>
>> Relevant URL(S):
>> For more information please contact: Mary McDavitt, 617-253-9620,
>> mmcdavit@csail.mit.edu
>>
>> _______________________________________________
>> Seminars mailing list
>> Seminars@lists.csail.mit.edu
>> https://lists.csail.mit.edu/mailman/listinfo/seminars
>>
>>
>>
>>
>
Re: [Fwd: TALK:Wednesday 10-26-05 End-to-End Performance Optimization of Java Server]
Posted by Davanum Srinivas <da...@gmail.com>.
I may be able to make it as well.
-- dims
On 10/18/05, Geir Magnusson Jr. <ge...@apache.org> wrote:
> Hey, I can be there, it seems. Who else? We can get together for
> lunch beforehand...
>
> geir
>
> On Oct 17, 2005, at 10:37 PM, Stefano Mazzocchi wrote:
>
> > for those in cambridge, ma or surroundings.
> >
> > --
> > Stefano.
> >
> >
> > From: CSAIL Event Calendar <ev...@csail.mit.edu>
> > Date: October 17, 2005 5:23:50 PM BDT
> > To: seminars@csail.mit.edu
> > Subject: TALK:Wednesday 10-26-05 End-to-End Performance
> > Optimization of Java Server
> >
> >
> >
> > End-to-End Performance Optimization of Java Server Workloads
> > Speaker: Jong-Deok Choi
> > Speaker Affiliation: IBM T. J. Watson Research Center
> > Host: Martin Rinard
> > Host Affiliation: CSAIL
> >
> > Date: 10-26-2005
> > Time: 2:00 PM - 3:00 PM
> > Refreshments: 1:45 PM
> > Location: 32-D463 Star
> >
> > The size and complexity of large-scale commercial middleware
> > systems, such as J2EE, and their workloads makes it difficult to
> > understand and optimize the performance of such systems. A J2EE
> > middleware with its application may have tens of thousands of
> > methods, hundreds of concurrently executing threads, and runtime
> > stacks that are hundreds of methods deep. While acting like an
> > operating system for its applications and providing various runtime
> > services, J2EE runs on a Java Virtual Machine (JVM) as an
> > application mostly written in Java. The JVM itself provides its own
> > various runtime services, such as synchronization and garbage
> > collection, and in turn relies on the underlying operating system
> > for OS services such as threading and memory management. This
> > complicated interaction among the various layers of the software
> > (and hardware) stack is a major source of the challenge to
> > understanding and optimizing the performance of such large-scale
> > middleware systems and their applicat
> > ions.
> >
> > In this talk, I will first present a brief description of the whole-
> > stack, end-to-end analysis and optimization system we have
> > developed at IBM T. J. Watson Research Center for Websphere
> > Application Server (WAS), which is IBM's implementation of the
> > J2EE. After that, I will present runtime data and their analysis
> > results from the various layers of the SW and HW stack running WAS
> > and its key benchmark programs, and compare them with those of
> > SPECjbb, a Java benchmark program widely used by Java performance
> > researchers. I will also present a few of the performance
> > bottlenecks we have found, and their optimization.
> >
> > Bio:
> > Jong-Deok Choi is a Research Staff Member (RSM) and Manager of the
> > High Performance Programming group at the IBM T. J. Watson Research
> > Center in Yorktown Heights, New York. He is a project co-leader of
> > the End-to-End Server Optimization project, a cross-organizational
> > effort in IBM for improving the performance of commercial eServer
> > applications on IBM's server platforms. Since joining Watson
> > Research in 1989, he has been involved in various research and
> > development projects - in the areas of program analysis,
> > optimization, and debugging - such as PTRAN, TPO, Jalapeno, DejaVu,
> > JikesRVM, and End-to-End Optimization. His research interests
> > include static and dynamic analysis and optimization of commercial
> > and scientific applications, middleware analysis and optimization,
> > high performance computer architecture, and debugging multithreaded
> > and distributed applications.
> >
> > He received B.S. in Electronic Engineering from SNU, Korea in 1979,
> > M.S. in Electrical Engineering from KAIST, Korea in 1981, and M.S.
> > and Ph.D. both in Computer Sciences from Univ. of Wisconsin-Madison
> > in 1985 and 1989, respectively.
> >
> > Relevant URL(S):
> > For more information please contact: Mary McDavitt, 617-253-9620,
> > mmcdavit@csail.mit.edu
> >
> > _______________________________________________
> > Seminars mailing list
> > Seminars@lists.csail.mit.edu
> > https://lists.csail.mit.edu/mailman/listinfo/seminars
> >
> >
> >
> >
>
> --
> Geir Magnusson Jr +1-203-665-6437
> geirm@apache.org
>
>
>
--
Davanum Srinivas : http://wso2.com/blogs/
Re: [Fwd: TALK:Wednesday 10-26-05 End-to-End Performance Optimization
of Java Server]
Posted by Stefano Mazzocchi <st...@apache.org>.
Geir Magnusson Jr. wrote:
> Hey, I can be there, it seems. Who else? We can get together for
> lunch beforehand...
awesome!
> geir
>
> On Oct 17, 2005, at 10:37 PM, Stefano Mazzocchi wrote:
>
>> for those in cambridge, ma or surroundings.
>>
>> --
>> Stefano.
>>
>>
>> From: CSAIL Event Calendar <ev...@csail.mit.edu>
>> Date: October 17, 2005 5:23:50 PM BDT
>> To: seminars@csail.mit.edu
>> Subject: TALK:Wednesday 10-26-05 End-to-End Performance Optimization
>> of Java Server
>>
>>
>>
>> End-to-End Performance Optimization of Java Server Workloads
>> Speaker: Jong-Deok Choi
>> Speaker Affiliation: IBM T. J. Watson Research Center
>> Host: Martin Rinard
>> Host Affiliation: CSAIL
>>
>> Date: 10-26-2005
>> Time: 2:00 PM - 3:00 PM
>> Refreshments: 1:45 PM
>> Location: 32-D463 Star
>>
>> The size and complexity of large-scale commercial middleware systems,
>> such as J2EE, and their workloads makes it difficult to understand
>> and optimize the performance of such systems. A J2EE middleware with
>> its application may have tens of thousands of methods, hundreds of
>> concurrently executing threads, and runtime stacks that are hundreds
>> of methods deep. While acting like an operating system for its
>> applications and providing various runtime services, J2EE runs on a
>> Java Virtual Machine (JVM) as an application mostly written in Java.
>> The JVM itself provides its own various runtime services, such as
>> synchronization and garbage collection, and in turn relies on the
>> underlying operating system for OS services such as threading and
>> memory management. This complicated interaction among the various
>> layers of the software (and hardware) stack is a major source of the
>> challenge to understanding and optimizing the performance of such
>> large-scale middleware systems and their applicat
>> ions.
>>
>> In this talk, I will first present a brief description of the whole-
>> stack, end-to-end analysis and optimization system we have developed
>> at IBM T. J. Watson Research Center for Websphere Application Server
>> (WAS), which is IBM's implementation of the J2EE. After that, I will
>> present runtime data and their analysis results from the various
>> layers of the SW and HW stack running WAS and its key benchmark
>> programs, and compare them with those of SPECjbb, a Java benchmark
>> program widely used by Java performance researchers. I will also
>> present a few of the performance bottlenecks we have found, and their
>> optimization.
>>
>> Bio:
>> Jong-Deok Choi is a Research Staff Member (RSM) and Manager of the
>> High Performance Programming group at the IBM T. J. Watson Research
>> Center in Yorktown Heights, New York. He is a project co-leader of
>> the End-to-End Server Optimization project, a cross-organizational
>> effort in IBM for improving the performance of commercial eServer
>> applications on IBM's server platforms. Since joining Watson Research
>> in 1989, he has been involved in various research and development
>> projects - in the areas of program analysis, optimization, and
>> debugging - such as PTRAN, TPO, Jalapeno, DejaVu, JikesRVM, and
>> End-to-End Optimization. His research interests include static and
>> dynamic analysis and optimization of commercial and scientific
>> applications, middleware analysis and optimization, high performance
>> computer architecture, and debugging multithreaded and distributed
>> applications.
>>
>> He received B.S. in Electronic Engineering from SNU, Korea in 1979,
>> M.S. in Electrical Engineering from KAIST, Korea in 1981, and M.S.
>> and Ph.D. both in Computer Sciences from Univ. of Wisconsin-Madison
>> in 1985 and 1989, respectively.
>>
>> Relevant URL(S):
>> For more information please contact: Mary McDavitt, 617-253-9620,
>> mmcdavit@csail.mit.edu
>>
>> _______________________________________________
>> Seminars mailing list
>> Seminars@lists.csail.mit.edu
>> https://lists.csail.mit.edu/mailman/listinfo/seminars
>>
>>
>>
>>
>
--
Stefano.
Re: [Fwd: TALK:Wednesday 10-26-05 End-to-End Performance Optimization of Java Server]
Posted by "Geir Magnusson Jr." <ge...@apache.org>.
Hey, I can be there, it seems. Who else? We can get together for
lunch beforehand...
geir
On Oct 17, 2005, at 10:37 PM, Stefano Mazzocchi wrote:
> for those in cambridge, ma or surroundings.
>
> --
> Stefano.
>
>
> From: CSAIL Event Calendar <ev...@csail.mit.edu>
> Date: October 17, 2005 5:23:50 PM BDT
> To: seminars@csail.mit.edu
> Subject: TALK:Wednesday 10-26-05 End-to-End Performance
> Optimization of Java Server
>
>
>
> End-to-End Performance Optimization of Java Server Workloads
> Speaker: Jong-Deok Choi
> Speaker Affiliation: IBM T. J. Watson Research Center
> Host: Martin Rinard
> Host Affiliation: CSAIL
>
> Date: 10-26-2005
> Time: 2:00 PM - 3:00 PM
> Refreshments: 1:45 PM
> Location: 32-D463 Star
>
> The size and complexity of large-scale commercial middleware
> systems, such as J2EE, and their workloads makes it difficult to
> understand and optimize the performance of such systems. A J2EE
> middleware with its application may have tens of thousands of
> methods, hundreds of concurrently executing threads, and runtime
> stacks that are hundreds of methods deep. While acting like an
> operating system for its applications and providing various runtime
> services, J2EE runs on a Java Virtual Machine (JVM) as an
> application mostly written in Java. The JVM itself provides its own
> various runtime services, such as synchronization and garbage
> collection, and in turn relies on the underlying operating system
> for OS services such as threading and memory management. This
> complicated interaction among the various layers of the software
> (and hardware) stack is a major source of the challenge to
> understanding and optimizing the performance of such large-scale
> middleware systems and their applicat
> ions.
>
> In this talk, I will first present a brief description of the whole-
> stack, end-to-end analysis and optimization system we have
> developed at IBM T. J. Watson Research Center for Websphere
> Application Server (WAS), which is IBM's implementation of the
> J2EE. After that, I will present runtime data and their analysis
> results from the various layers of the SW and HW stack running WAS
> and its key benchmark programs, and compare them with those of
> SPECjbb, a Java benchmark program widely used by Java performance
> researchers. I will also present a few of the performance
> bottlenecks we have found, and their optimization.
>
> Bio:
> Jong-Deok Choi is a Research Staff Member (RSM) and Manager of the
> High Performance Programming group at the IBM T. J. Watson Research
> Center in Yorktown Heights, New York. He is a project co-leader of
> the End-to-End Server Optimization project, a cross-organizational
> effort in IBM for improving the performance of commercial eServer
> applications on IBM's server platforms. Since joining Watson
> Research in 1989, he has been involved in various research and
> development projects - in the areas of program analysis,
> optimization, and debugging - such as PTRAN, TPO, Jalapeno, DejaVu,
> JikesRVM, and End-to-End Optimization. His research interests
> include static and dynamic analysis and optimization of commercial
> and scientific applications, middleware analysis and optimization,
> high performance computer architecture, and debugging multithreaded
> and distributed applications.
>
> He received B.S. in Electronic Engineering from SNU, Korea in 1979,
> M.S. in Electrical Engineering from KAIST, Korea in 1981, and M.S.
> and Ph.D. both in Computer Sciences from Univ. of Wisconsin-Madison
> in 1985 and 1989, respectively.
>
> Relevant URL(S):
> For more information please contact: Mary McDavitt, 617-253-9620,
> mmcdavit@csail.mit.edu
>
> _______________________________________________
> Seminars mailing list
> Seminars@lists.csail.mit.edu
> https://lists.csail.mit.edu/mailman/listinfo/seminars
>
>
>
>
--
Geir Magnusson Jr +1-203-665-6437
geirm@apache.org