You may have already seen how fast the WSO2 Enterprise service Bus (ESB) is. Want to learn how to use this ultra fast, light-weight and versatile Enterprise Service Bus (ESB) powered by the Apache Synapse ESB?
Well we got an answer for you. Join our free webinar series this July. There will be four webinars in total ranging from an "Introduction to the WSO2 ESB" to more advanced scenarios such as "Using the WSO2 ESB and FIX - supporting Financial messaging".
FIX support is the latest addition to the WSO2 ESB which is available in the recently released 1.7 version. This series of webinars will be presented by Paul Fremantle, CTO at WSO2 together with Asankha Perera ( ESB Architect), Asanka Abeysinghe (Architect) and Ruwan Linton (Senior Software Engineer).
Saturday, June 28, 2008
Friday, June 27, 2008
Increasing memory of JUnit TestCases in Maven2
The WSO2 Mashup Server uses Maven2 as its build system. Since of late we were facing an issue where our integration tests were failing regularly with the exception "Caused by: java.lang.OutOfMemoryError: Java heap space". This is the JIRA that tracked this issue . During the integration test phase we start up an instance of the mashup server and deploy some test scripts. We them bombard the server with around 110 request (within a second or two) hence it ran out of memory before the tests completed.
Increasing memory given to maven using MAVEN_OPTS did not solve the problem. As this setting just increases the memory given to the build process and not the individual test cases. The solution was to use the following configuration in the surefire plugin.
The pom for the integration test can be found here. This fixed the heap space issue and now we have our test cases running on the nightly build. You can see the status of our nightly builds at http://builder.wso2.org/browse/MASHUP-NIGHTLY
Increasing memory given to maven using MAVEN_OPTS did not solve the problem. As this setting just increases the memory given to the build process and not the individual test cases. The solution was to use the following configuration in the surefire plugin.
<plugin>
<groupId>org.apache.maven.plugins</groupId>
<artifactId>maven-surefire-plugin</artifactId>
<configuration>
<forkMode>pertest</forkMode>
<argLine>-Xms256m -Xmx512m</argLine> <testFailureIgnore>false</testFailureIgnore>
<skip>false</skip>
<includes>
<include>**/*IntegrationTestSuite.java</include>
</includes>
</configuration>
</plugin>
The pom for the integration test can be found here. This fixed the heap space issue and now we have our test cases running on the nightly build. You can see the status of our nightly builds at http://builder.wso2.org/browse/MASHUP-NIGHTLY
Labels:
Integration,
JUnit,
Mashup Server,
Maven2,
Surefire plugin
Is Mule Really Open Source?
Most Open Source Projects (Not the namesake open source projects) are released under an OSI approved license. The open source definition of OSI states that "Open source doesn't just mean access to the source code. The distribution terms of open-source software must comply with the following criteria:". It goes on to describe 10 items under this. Item #6 states that "The license must not restrict anyone from making use of the program in a specific field of endeavor. For example, it may not restrict the program from being used in a business, or from being used for genetic research."
The Mule Enterprise Edition (Mule EE) violates item #6 which is described above. The License of Mule EE states that you cannot publicly publish performance results against it. Whats more if you go to the mule site you won't find the source code of Mule EE, rather it asks you weather you are "Looking for the open source Mule Community Edition?".
What baffles me is that Dave Rosenberg the CEO Mule states that "The Enterprise version of Mule is open source". Take a look at Dave's comment on Sanjiva's blog. What a joke.
Now Mule CE is an open source ESB. But would anybody care to use it in production? Look what we found out about Mule CE during some performance testing. 1% of all requests sent to Mule CE 2.0.1 fails. Now thats how you sell Mule EE keeping Mule CE Open Source.
The Mule Enterprise Edition (Mule EE) violates item #6 which is described above. The License of Mule EE states that you cannot publicly publish performance results against it. Whats more if you go to the mule site you won't find the source code of Mule EE, rather it asks you weather you are "Looking for the open source Mule Community Edition?".
What baffles me is that Dave Rosenberg the CEO Mule states that "The Enterprise version of Mule is open source". Take a look at Dave's comment on Sanjiva's blog. What a joke.
Now Mule CE is an open source ESB. But would anybody care to use it in production? Look what we found out about Mule CE during some performance testing. 1% of all requests sent to Mule CE 2.0.1 fails. Now thats how you sell Mule EE keeping Mule CE Open Source.
Training session on Apache Axis2
A couple of weeks ago we were fortunate to attend a training on Axis2 which was conducted by Deepal. There is no better person to teach Axis2 than Deepal himself cause he has been involved with this project since its inception. Deepal is one of the six core Axis2 authors. The others are Srinath, Eran Chinthaka, Ajith, Chathura and Jaliya. There have been many others who've help Axis2 once the project started going, but these were the guys involved in building the core of it. Not to forget Dr. Sanjiva for all the advice and guidance provided.
The guys who knew Axis2 to some extent were able to make to most of this training, as we were able to question certain decisions made at that time.
Its also nice to note that except for Deepal. the rest are already reading for there PhDs. Deepal will be joining that list next month when he will head off for higher studies. Its also a interesting that the core Axis2 team has been associated with WSO2.
Deepal has also published his first book on Axis2 "Quickstart Apache Axis2" which is now available in the stores.
The Axis2 training was a great success so now its time to learn some policy. Again there will be no better person than Sanka himself, the person behind Apache Neethi (The policy engine used for Apache Axis2 as well as CXF). Once next week is done all of WSO2 will be policy experts too.
All of these are a part on continuous improvement to us.
The guys who knew Axis2 to some extent were able to make to most of this training, as we were able to question certain decisions made at that time.
Its also nice to note that except for Deepal. the rest are already reading for there PhDs. Deepal will be joining that list next month when he will head off for higher studies. Its also a interesting that the core Axis2 team has been associated with WSO2.
Deepal has also published his first book on Axis2 "Quickstart Apache Axis2" which is now available in the stores.
The Axis2 training was a great success so now its time to learn some policy. Again there will be no better person than Sanka himself, the person behind Apache Neethi (The policy engine used for Apache Axis2 as well as CXF). Once next week is done all of WSO2 will be policy experts too.
All of these are a part on continuous improvement to us.
Thursday, June 26, 2008
Enterprise Service Bus (ESB) Performance Revealed
It was recently that WSO2 released the 1.7 version of its Enterprise Service Bus (ESB). This release is based on Apache Synapse 1.2 which in turn is built on Apache Axis2 1.4.
There are many things that come in to play when deciding on which ESB to go with. Some of them are Performance, Licensing, Usability, Stability... etc. WSO2 has always been at the forefront when it comes to benchmarking ESB performance. You can find previous performance comparisons here and here. We've always been open as to how we conduct them and has always tried to get help from the relevant projects on how to configure them best.
Asankha the product manager of the WSO2 Enterprise Service Bus (ESB) recently published performance results based on its 1.7 release. The WSO2 Enterprise Service Bus (ESB) was compared with other leading implementations - both open source and proprietary. This comparison was done against a leading proprietary ESB, the proprietary version of an Open Source ESB, Apache Service Mix 3.2.1 and Mule Community Edition 2.0.1 (Mule CE). We cannot disclose the names of these proprietary ESB's cause they have a clause in there License saying that "Thou shalt not publish any performance results against us".
The results were pretty impressive. The WSO2 ESB outperformed all open source alternatives and it also outperformed the proprietary version of an Open Source ESB. Now what does that mean? The WSO2 ESB is the fastest open source ESB around (And thats with no specific performance tuning, Expect better results in the time to come.). The WSO2 Enterprise Service Bus (ESB) also displayed excellent stability on all scenarios.
Another interesting observation is that 1% of all requests sent to Mule CE 2.0.1 failing. Now who would want to use such a product in a production environment?
There are many things that come in to play when deciding on which ESB to go with. Some of them are Performance, Licensing, Usability, Stability... etc. WSO2 has always been at the forefront when it comes to benchmarking ESB performance. You can find previous performance comparisons here and here. We've always been open as to how we conduct them and has always tried to get help from the relevant projects on how to configure them best.
Asankha the product manager of the WSO2 Enterprise Service Bus (ESB) recently published performance results based on its 1.7 release. The WSO2 Enterprise Service Bus (ESB) was compared with other leading implementations - both open source and proprietary. This comparison was done against a leading proprietary ESB, the proprietary version of an Open Source ESB, Apache Service Mix 3.2.1 and Mule Community Edition 2.0.1 (Mule CE). We cannot disclose the names of these proprietary ESB's cause they have a clause in there License saying that "Thou shalt not publish any performance results against us".
The results were pretty impressive. The WSO2 ESB outperformed all open source alternatives and it also outperformed the proprietary version of an Open Source ESB. Now what does that mean? The WSO2 ESB is the fastest open source ESB around (And thats with no specific performance tuning, Expect better results in the time to come.). The WSO2 Enterprise Service Bus (ESB) also displayed excellent stability on all scenarios.
Another interesting observation is that 1% of all requests sent to Mule CE 2.0.1 failing. Now who would want to use such a product in a production environment?
Labels:
Enterprice Service Bus,
ESB,
Mule,
Performance,
WSO2
Friday, June 6, 2008
Paul among the Best
Paul has been listed as one of Infoworld's top 25 CTOs. This is a huge achievement for Paul as well as WSO2 (whom Paul is a co-founder of).
Take a look at what Paul himself has to say about this achievement. I agree with Paul a 100% about what he has to say about the culture at WSO2, "I strongly believe - and long may it continue - that WSO2 doesn't suffer from "Drive By Architecture" syndrome". WSO2 is flat structured, even though there are co-founders, CTOs, Product Managers and so on when it comes to a technical debate each and every member at WSO2 has a role to play. Their thoughts and ideas are taken seriously and is not looked down on.
The team spirit we have at WSO2 is tremendous and it's a pleasure to be a part of WSO2.
Thursday, June 5, 2008
University of Moratuwa ranked first at GSoC 2008
Its really encouraging to see how much open source has cought up in Sri Lanka. A lot if the credit in this regard should go to visionaries like Dr. Sanjiva Weerawarana who initiated the Lanka Software Foundation (LSF) and later on went on to founder WSO2.
A living testimony to the development of open source in Sri Lanka is GSoC 2008. Where Sri Lanka was ranked 9th in the list of GSoc Applications and GSoc Accepted Applications. Not surprisingly a staggering number of applications from Sri Lanka came from the University of Moratuwa which is where I was privileged to have had my Higher education at. The University of Moratuwa ranked 1st in both the number of Applicants and number of accepted applications.
A living testimony to the development of open source in Sri Lanka is GSoC 2008. Where Sri Lanka was ranked 9th in the list of GSoc Applications and GSoc Accepted Applications. Not surprisingly a staggering number of applications from Sri Lanka came from the University of Moratuwa which is where I was privileged to have had my Higher education at. The University of Moratuwa ranked 1st in both the number of Applicants and number of accepted applications.
Subscribe to:
Posts (Atom)