Tuesday Feb 10, 2009

Updating WebSphere, Java and Solaris for Better Performance and Reliability

Keeping your WebSphere application environment on Solaris updated with the recommended Fix Packs and Service Releases from IBM and Solaris update releases and patches is a key to better performance and reliability. For example, if you have WebSphere Application Server v6.1 with Fix Pack 5, you may find slowness in the garbage collection process using the throughput collector (ParallelGC).[Read More]

Monday Oct 06, 2008

WebSphere Application Server and JVM Options

For WebSphere Application Server (WAS) on Solaris, IBM bundles it with the Sun JVM that contains IBM's modifications (e.g. IBM ORB, XML and Security libraries) as explained in this blog entry by my colleague Dileep Kumar of Sun.

More specifically, you can find the summary of WAS and its Java versions in the table below. Additional details can be located at this IBM site.

WAS v7.0Java EE 5JDK/JRE 6
WAS v6.1J2EE 1.4JDK/JRE 5
WAS v6.0.2J2EE 1.4JDK/JRE 1.4.2
WAS v5.1.1J2EE 1.3JDK/JRE 1.4.2
WAS v5.1J2EE 1.3JDK/JRE 1.3.x

In order to tune WAS JVM on Solaris, it is greatly helpful if you understand the performance tuning aspects of Sun JVM such as Java memory management, Garbage Collectors, etc. You can get performance whitepapers for respective JVM versions here and Java Memory Management Whitepaper. Dileep also has blogged about specific recommended JVM options for WAS on Solaris here.

Finally, there are numerous Sun JVM options and one of the references that I strongly recommend is the collection compiled by Joseph D. Mocker of Sun Microsystem. Of course, we have much more specific information about WAS v6.1 on Solaris 10 in this IBM Redbook (e.g. Chapter 9).

Thursday Apr 17, 2008

New Sun SPARC Enterprise T5x40 servers for IBM WebSphere

While we were at the IBM Impact 2008 Conference, Sun announced the new UltraSPARC-T2 Plus servers. Here's a blog about the announcement. You can also learn more about it at this Sun CMT Wiki.

[Read More]

Wednesday Apr 16, 2008

Great meet in Las Vegas

We presented multiple technical sessions at Impact 2008 and certainly appreciated the attendance. We were asked if we'd repeat SIG-2730A, titled "Deployment Best Practices for WebSphere Application Server Version 6.1 on Solaris 10", and we did it on Thursday. We gave away many copies of our Redbook and glad to have received excellent feedbacks. And, we enjoyed autographing them, too.

I'd like to personally thank to all those who attended the sessions, came to meet with us at the "Solution Center", and requested us for individual meetings. We enjoyed discussing about your activities/challenges and how we could help address them. We are glad that we're able to provide value-add solutions for your WebSphere deployment.

I have made the presentations available at:

Thursday Mar 06, 2008

New Redbook Published: "WebSphere Application Server V6.1 on the Solaris 10 Operating System"

SG247584 Redbook Cover The final version of the Redbook titled "WebSphere Application Server V6.1 on the Solaris 10 Operating System" has been published. In this book, we document how WebSphere Application Server V6.1 and Solaris 10 can be configured, optimized, and managed. We described how to virtualize and manage WAS installation, deployment strategies and scenarios, the advanced features of Solaris 10 (e.g. SMF, Resource Management, Process Rights, Containers and Zones, ZFS, DTrace), the differences of WAS on Solaris from other platforms (e.g. Sun JDK), how to monitor and tune WAS along with Sun JVM, Solaris, and much more.

I personally would like to thank to all those from both Sun and IBM sides who helped make this project happened as well as those who participated and contributed on the project. I am very appreciative of those reviewers from various Sun and IBM engineering organizations who spent many hours for accuracy and correctness of the content. We tried our best to recognize them in the Preface of the book. If we inadvertently left any of you out, we apologize. I also am thankful for those who sent me great questions and feedback for the book from around the world.

Last but not the least, I must thank my family 'AAAA' for giving me love and support while I was working on the book many weeks and nights away from and at home. Without you, I couldn't have accomplished this. I love you all.

Finally, below are some comments we got so far for the book. Please feel free to send them along, rate it at the Redbook site, or comment here on the blog.

  • "Good informations, we use it a lot", "Excellent Redbook...very informative!"
  • " Good work.", "Great..We have been waiting for."
  • "Excellent", "Very useful, This Redbook is very useful for our Business Partners, because they need the information and technique how to configure/use WebSphere Application Server on Solaris 10 virtualization technology."

So, go download the following and make most use of your Solaris platform -- the best OS on the planet!


Tuesday Jan 29, 2008

Sun Modular Data Center

The project known as "Blackbox" is an official Sun product now. It's called the Sun Modular Data Center S20.
Sun Blackbox Figure
Sun Modular Data Center has already been in deployment around the world:
  • Stanford Linear Accelerator Center (SLAC), Menlo Park, CA
  • Mobile TeleSystems OJSC (MTS) Moscow, Russia
  • Hansen Transmissions' (manufacturing facility), India
  • Radboud University Nijmegen Medical Centre (UMCN), the Netherlands
Sun Modular Data Center offers cost-effective, high-density compute capacity, and flexibility. Its building blocks include many of the "volume" and blade systems that you can buy from Sun today. It's all about better design of Sun systems that enable such a data center.

Tuesday Jan 22, 2008

A New Redbook (DRAFT): WebSphere Application Server V6.1 on Solaris

Sun and IBM WebSphere teams have collaborated together to write this new Redbook titled "IBM WebSphere Application Server V6.1 on the Solaris Operating System". I, along with my Sun colleague Dileep Kumar, were part of the IBM Redbook Residency in November and December of 2007. We enjoyed working with the IBM team that included Phil Monson (one of the finest project managers), Dennis Martin, Ilari Ahtiainen and Joe Reinwald on the project.

The book is now available as a draft Redbook on the internet at: http://www.redbooks.ibm.com/redpieces/abstracts/sg247584.html.

This book documents the best practices of deployment and performance management of WAS v6.1 on Solaris 10. Chapters include:

  • Chapter 1. Introduction to WebSphere Application Server on Sun Solaris
  • Chapter 2. Configuring a Solaris environment for WebSphere Application Server
  • Chapter 3. Installation of WebSphere Application Server for Solaris 10
  • Chapter 4. Installation of WebSphere Application Server Network Deployment for Solaris 10
  • Chapter 5. Configuration of WebSphere Application Server in Advanced Solaris 10 Environment
  • Chapter 6. Management and Maintenance of WAS on Solaris
  • Chapter 7. Advanced Topologies
  • Chapter 8. Security and Identity Management
  • Chapter 9. WebSphere Application Server Performance Tuning on Solaris 10
  • Chapter 10. Problem Determination
  • Appendix A. WebSphere Application Server V6.0.2 considerations
  • Appendix B. Sun Solaris 8 Migration Assistant (Etude)
  • Appendix C. Additional material

We are still ironing out some contents and quite close to a release version. Please provide us with your feedback. You can reach me at {FirstName}.{LastName}@Sun.COM -- see my name at the top of my blog.

Friday Jan 04, 2008

Clarifying WebSphere XD's ODR Thread Limitation on CMT systems

In the IBM WebSphere eXtended Deployment (XD) On-Demand Router (ODR), the default thread pool size is set to a small value (i.e. 4 threads). The customers who use systems that have many hardware threads (e.g. Chip Multi-Thread (CMT) aka CoolThreads US-T1 and US-T2 processor based systems) perceive or are led to perceive that this is the limitation on these Sun CoolThreads system.

That is not true. The default thread pool size in ODR is configurable.

However, you do not want to make the default pool size to be too large (e.g. US-T2 has max 64 threads) as ODR does not need many threads to execute because of its asynchronous nature. It is more efficient to run the ODR with fewer threads to avoid getting into unnecessary context switching.


Mostly pertaining to Cloud Computing, Application Infrastructure, Oracle Exastack, Exalogic, Solaris, Java and Sun servers for the enterprise!


« May 2016