Thursday Apr 30, 2009

IMPACT 2009 @ Las Vegas

Wanted to let you all know that Sun/IBM Team will be hosting the following BOF during this year IMPACT 2009.
Title: Optimizing your IBM WebSphere® Infrastructure with Solaris
Session ID: BOF-3174A
Session Time: Tue, May 5 12:00 PM - 01:00 PM
Location: Venetian Galileo 904
This will give you an unique opportunity to talk to engineers from both Sun and IBM team. We can cover whole range of topics concerning your WebSphere deployment on Sun platform.

For rest of the discussion I suggest you stop by at our booth and we can help you get your question answered.

Tuesday Apr 14, 2009

Scaling up WebSphere on Sun Blade X6270 Server

As always I got chance to try my hands on one of the new blade server that Sun has released based on Intel Xeon processor 5500 series namely, Sun Blade X6270 server module. The new Xeon processor has lot of features to deliver better performance based on the workload by which it is driven eg. automatically increasing processor frequency when needed and possible, as well as the capability to have extra hardware threads or taking advantage of the processor's power and thermal headroom. I would suggest you to visit for more details from domain expert including power consumption to virtualization and everything else supplemented by best OS on the planet "Solaris 10/OpenSolaris". I will leave rest of the discussion to my distinguished colleagues at Sun and will talk about what I observed when running WebSphere Application Server on this system.

As most current version of WebSphere Application Server available was V7.0 so I picked up this installed then made sure I am running with the latest fixpacks and ran it. Everything was breeze from installation, fix-pack upgrade to start/stop and then benchmarking. The Sun Blade X6270 I had was built on 2 sockets of the Intel Xeon L5520 processor and the Solaris 10 10/08 Operating System.

As always I used the same workload that I use for my WebSphere tests and there are other components which included the Database Server which was running DB2, and the test Drivers. For benchmark I used Faban which is very easy to use and run and it does the bookkeeping of benchmark so nicely that I don't have to keep records myself. I don't constrain my DB in terms of performance so I let it run in "/tmp" to just make everything simple so I don't spend too much time on tuning DB as I am working on WebSphere App Server.

The Intel Xeon processor 5500 series has two modes for running namely, Turbo mode and Hyper-Threading mode options in the BIOS setup. Which can be changed during boot by going to BIOS menu and doing the right selection. As Intel has put up "Nehalem, unleashes parallel processing performance enabled by an integrated memory controller and Intel® QuickPath Technology providing high-speed interconnects per independent processing core. The Intel® Turbo Boost Technology taking advantage of the processor's power and thermal headroom. This enables increased performance of both multi-threaded and single-threaded workloads and with Intel HT Technology, bringing high-performance applications into mainstream computing with 1-16+ threads optimized for new generation multi-core processor architecture." I did tried to run in both modes did not notice any significant difference for my tests while running in these modes.

Single instance of server was able to use much of the system capacity and stopped scaling. I was running all the tools and nicstat was one able to point me the moment when the n/w bandwidth was exhausted it stopped taking any more load. Then to work around the network capacity I put a direct connection with the DB Server which resulted in little better throughput and little more system utilization. On the network communication between DB and Server was happening over this interface. Still not enough and server had capacity neither the thread-dumps or system stats pointed out any problem other than network again being saturated. In which the public interface was full loaded and there was a very light load on the point to point connection interface. At this point I don't had any choice other than adding another instance of the server. To do this I created a Solaris Zone/Container and had one instance running into this zone. For this zone I used the point to point connection between the DB and the blade server which eliminated this and I was able to use the server to full capacity.

The network generated interrupts and there distribution can be checked by using the following command:
-bash-3.00# echo ::interrupts | mdb -k | grep igb
49 0x60 6 MSI-X 0 1 - igb_intr_tx_other
50 0x61 6 MSI-X 0 1 - igb_intr_rx
51 0x62 6 MSI-X 4 1 - igb_intr_tx_other
52 0x63 6 MSI-X 7 1 - igb_intr_rx

I looked for igb only becuase my network drivers are as:
igb0 - public (ie. going over GB switch)
igb1 - private (point-to-point)
From the above it is clear that the tx and rx interrupts are going over virtual cpu or more technically threads 0, 4 and 7 which is very well distributed.

So what did I do for tuning the WebSphere ?
My tunings remains same as I was not changing the Operating System and Software which includes Java. I will list them here for reference:

Disabled the Performance Monitoring Infrastructure
Disabled the Application Profiling
I used the DynaCache for the App just as to save some DB overhead
Set the WebContainer thread to the same number of threads as load driver
For Connection Pool I made sure that the number of DB connections are more than my WebContainer thread pool so none of the thread end up waiting for DB connections.
initialHeapSize="2048" maximumHeapSize="2048" -server -Xmn1024m -XX:+AggressiveOpts -XX:+UseParallelGC -XX:ParallelGCThreads=8 -XX:+UseParallelOldGC

Before I conclude I would like you to know that the only 64-bit of WebSphere is available on this platform so if your application happen to be memory hungry you are not limited by a 4GB address space and as you notice from David Dagastine blog that there are lot of things happening with 64-bit JVM and how it is getting optimized compared to 32-bit variant. But you may have to wait for the appropriate fixpacks to get to that. I will update as soon as it becomes available.
There are lot of other advantages of this server and I would encourage you to read more about this @

Monday Mar 23, 2009

Free cryptography for WebSphere on Ultra SPARC T2 and T2 plus systems

If you have been being hit by the processing overhead of SSL stuff and have to buy special purpose hardware to speed up things, then may be its time to leave all this processing to Ultra SPARC T2 and T2 Plus based system. Which has builtin cryptogarphic support on the chip and websphere supports this:

Here are the details from IBM WebSite:

For details of setup you can follow the WebSphere Infocenter.
Please note that this feature will be available only after certain fix pack release so it will not work unless you upgrade your JDK to required level as suggested in the doc.

If you want to learn more about the Cryptographic Acceleration offered by these processor I would encourage to read this Sun Blueprint titled "Using the Cryptographic Accelerators in the UltraSPARC® T1 and T2 Processors".

Ask the expert session for WebSphere on Solaris

We will be hosting an Ask The Experts session on March 24th. 
We invite you all to join where we can discuss and share various ideas around how to optimally use WebSphere software on Solaris platform.
For  details please visit the WebSphere User Group web-site(

Thursday Jan 08, 2009

WebSphere Application Server V7.0

By this time you might have got your hand on new WebSphere Application Server namely WebSphere Application Server v7.0. There are lot of good thing about this version as compared to its predecessors. A list of which can be found easily from 7.0 InfoCenter.
Some of the feature as Solaris user you should be aware are:

  • New JDK/JVM Support

  • DTrace made easy

  • Performance Enhancement

  • New J2EE Spec Level Compliance

All the detail can be found at
Some of the differences you can note from the standard Java Installation with the one that is part of WAS 7 s missing sample directory which contains lot of samples including DTrace scripts which are ready to be used. If you want to try them grab a JDK 6.X installer. Install it and copy the dir YOUR_JAVA_INSTALL_DIR/sample/dtrace and start using the script as it is. There is a README.txt file which included in the dtrace dir. All the sample scripts also has instruction about how to use them. More information about DTrace can be found here.

Monday Oct 13, 2008

WebSphere Application Server and T5440

Today, we are announcing the next generation server based on UltraSPARC T2 Plus processor and this server happens to be a monster in terms of performance. It can consolidate a lot of servers in a single box providing the similar combined throughput. For the server details and other blogs related to this server I suggest you to visit T5440 Blog index(
In this blog, I summarize the performance of IBM WebSphere Application Server (WAS)  on T5440. To test and benchmark the performance, I use the most recent release WAS v7, that is based on Java EE 5 spec and JDK 6, with numerous features and performance enhancements. As I said earlier that this new Sun server is a monster in terms of performance, the WAS software set up and configuration require some planning and appropriate allocation of the system resources.   When you do the psrinfo(1M) command on this server, it will report to you that the system has 256 "processors".  This means that the processing power of this system far exceeds the software scalability of a single instance of the Application Server's. Thus, you will need multiple instances of WAS to drive the system to its full utilization.  Solaris Containers provides the most efficient way to accomplish such configuration as Solaris Containers provides process space isolation among different WAS instance as well as allocating proper system resources.
For maximizing the utilization of the system, I configured the environment as follows.

  • I created 7 Solaris Containers, allocated 32 processor threads to each of the 6 Containers, and allocated the remaining processor threads for the other container and the global zone.  Then, I used the following WAS configuration:

  • initialHeapSize="2500" maximumHeapSize="2500"

  • -server -Xmn2048m -XX:+AggressiveOpts -XX:+UseParallelGC -XX:ParallelGCThreads=16 -XX:+UseParallelOldGC -Djavax.xml.transform.TransformerFactory=org.apache.xalan.processor.TransformerFactoryImpl -Djavax.xml.parsers.SAXParserFactory=org.apache.xerces.jaxp.SAXParserFactoryImpl -Djavax.xml.parsers.DocumentBuilderFactory=org.apache.xerces.jaxp.DocumentBuilderFactoryImpl -Dorg.apache.xerces.xni.parser.XMLParserConfiguration=org.apache.xerces.parsers.XML11Configuration

  • Disable the PMI feature.

  • Disable the System.out logging from the admin console if you want you can do the same by editing the config file. by editing the trace service like - startupTraceSpecification="\*=info:SystemOut=off"

  • DynaCache - The DynaCache was used for this benchmark so the default size of the cahce was set to 2000 which wasn't enough for this test which can be adjusted accordingly based on the application need as I set it to 3000.

  • For thread pool tuning the only thing that was relevant in this benchmark was tuning the WebContainer thread pool right so I did set it to 80 and which was more that enough and I think It can be scaled down a bit but havn't tried.

  • Database connection pool was set it to same value as WebContainer pool and set the min and max to the same value.

For database I did not want to run into the network/disk contention so I created the DB in /tmp and I had to use two databases for this purpose. This was just to eliminated the Database configuration headaches and just measure the App Server machine scalability. These database were connected point-to-point with App Server box and the App Server instances were using the point-to-point connection for the DB.
This resulted in getting to 1.8X scalability of its predecessor system(T5140) in terms of throughput(which is measured for this benchmark in terms of requests served per second or more commonly known as req/sec).
So In nutshell if you are looking for another system to consolidate your WebSphere deployment with 1.8X throughput capacity of earlier UltraSparc T2 plus system then T5440 may be just the right system for you.

Wednesday Apr 23, 2008

Thank you all for conversation during IMPACT 2008, Las Vegas

We would like to thank all those who stopped by and talked to us during IMPACT. It was a great conference and it was great to talk to you all about your Solaris 10 deployments and how well your deployment is running on Solaris 10.
There were few really great session and I would suggest you get a copy of some of the interesting presentation eg.
1. IBM WebSphere and multicore platforms: Performance, scalability, and best practices
2. Exploring IBM Java 5 and Java 6: New features and enhancements
3. Understanding and tuning the Java Virtual Machine for IBM WebSphere
for you reference. They contain lots of good information which you may found useful later.

Albert has already posted the slides which we used during our presentation. They can be found here:
SIG2730: WebSphere Deployment on Solaris
TSM-2447: WebSphere Performance On Solaris

There was one other session which was on DumpAnalyzer which is not available on Solaris or HP-UX I will put another blog which will list out all the equivalent tools that already available in the Java you have on your system with WebSphere.

Tuesday Mar 18, 2008

Coming to IMPACT 2008

This year IBM is having IMPACT 2008 event in Las Vegas, NV from 6 April 2008. If you are planning to be there then here are the few sessions which we are presenting you may find interesting. BOF is going to be very very interesting and we are planning to have a open discussion where we can go through whatever is most needed by customers/WAS admins for their WAS deployments on Solaris 10. If you are looking for some collaterals or white-papers which is not available today we can certainly take down your request, prioritize them and then make them available as soon as possible. So don't miss this opportunities to talk to us and let us know what are the information needed for worry free deployment of WAS on Solaris. The details of the sessions are as follows which we will be presenting :

1) Technical Session: Performance Best Practices for WebSphere Application Server Version 6.1 on Solaris 10
Session #: TSM-2447
Session Time: Thu, 10/Apr, 04:45 PM - 06:00 PM
Location: Room 118

2) BOF: Deployment Best Practices for WebSphere Application Server Version 6.1 on Solaris 10
Session #: SIG-2730A

3) IT Executive Session: Sun and IBM WebSphere Message Broker: Configuration, Deployment and tuning on Solaris 10
Session #: ITE-2695

Wednesday Jan 16, 2008

Redbook for WebSphere on Solaris

Last couple of week we don't had any update over here as I was working with some of the IBM folks(really great ones..) on a Redbook project. The redbook is titled something like "IBM WebSphere on Sun Solaris" and will be available in couple of weeks. Just wanted to let everyone know that this is going to be a onestop information for your Websphere deployment on Solaris. I am sure some of the chapters are really great and will answer may of the questions being asked frequently.

Here is the link to the internet draft:




« July 2016