Archive for the ‘Technology’ Category

ITCAM for SOA and WSRR

Friday, March 2nd, 2007

IBM Tivoli Composite Application Manager for SOA (or ITCAM for SOA as it is usually referred to) is, as the name suggests, an important part of IBM’s monitoring solution for the SOA, tracking web service requests not only through IBM products such as WebSphere Application Server, DataPower and WebSphere ESB but also into other environments like SAP NetWeaver and JBoss. WebSphere Service Registry and Repository has a number of integration points with ITCAM for SOA, one of which is an Event Handler. This enables situations detected by ITCAM for SOA (such as an excessive response time or message size) to result in the creation, update, or removal, of properties on a WSDL port or SCA export in the registry. You can read a developerWorks article that describes how to configure ITCAM for SOA to monitor a web service running in WebSphere Application Server and update the metadata for the service entry in the registry.

One possible usage is to modify the behaviour of clients depending on the current system status. For example, by using a query to select endpoints which are not currently experiencing response time issues. This isn’t currently possible in WebSphere ESB when using the endpoint lookup primitive due to the internal caching that is performed. Unlike WebSphere Message Broker, the primitive does not currently have any mechanism to invalidate the cache when service definitions are modified.

WebSphere ESB and Messaging

Friday, March 2nd, 2007

There are two new articles on developerWorks covering connectivity between WebSphere ESB and messaging. The first, is Part 4 of Rachel and Andre’s series which takes a look at the new Websphere MQ binding in Version 6.0.2. The second article shows how to use the JMS bindings with JMS providers other than the default messaging provider or WebSphere MQ i.e. what WebSphere Application Server refers to as generic JMS providers.

WS-Addressing endpoint references

Friday, March 2nd, 2007

Ben Bakowski has written a very thorough developerWorks article covering the use of WS-Addressing endpoint references with WebSphere Application Server Version 6.1 to act as references to stateful session beans. Ben uses his sample online shop application to show the possibilities including how to achieve high availability in a cluster.

WebSphere User Group meeting

Tuesday, February 27th, 2007

I headed down to Bedfont early this morning for the WebSphere User Group meeting. Such is the traffic on the M3 that I missed the start but arrived in time for Jim Caldwell’s keynote presentation. Jim is the IBM Director of WebSphere Application Infrastructure and had some interesting things to say about many parts of the portfolio from WAS CE to WebSphere XD. Two products were mentioned that I’ve never really paid much attention to in the past. WebSphere Real Time is a Java environment for real-time applications, providing for sub-second response times free from the usual vagaries of garbage collection. Meanwhile, WebSphere Remote Server is targetted at the retail market, providing a J2EE runtime for the store with remote management capabilities.

(more…)

Restricted access

Thursday, February 8th, 2007

My colleague Simon Kapadia informs that he can’t access this site from our standard issue 3G cards due to “Vodafone Content Control”. Fortunately if you’re over 18 you can remove the access control (which I’m guessing I must have done at some point in the past) but make sure the children aren’t watching!

More from developerWorks

Friday, February 2nd, 2007

I’ve been catching up on some of the developerWorks articles published in the past week and here are a selection of the best. First up is an up-date to the top Java EE best practices. As the article states, it’s amazing how many customers still aren’t following these simple steps. On the WebSphere ESB front we have the third part in Rachel and Andre’s series on Building an Enterprise Service Bus using WebSphere ESB. This looks at using SOAP/HTTP bindings, property promotion and administrative modification of endpoint addresses. Greg Flurry goes one step further in his article, covering the new dynamic endpoint capability in WebSphere ESB V6.0.2 including the use of the endpoint lookup primitive in combination with WebSphere Service Registry and Repository. This leads me on nicely to a new series which looks at the use of generic objects in Service Registry to group related documents.

Service Integration Bus Performance

Monday, January 29th, 2007

Another great tool for those working with the Service Integration Bus is now available on alphaWorks. The Service Integration Bus Performance tool (written by David Granshaw who leads the Service Integration Bus Performance team in Hursley) provides an SWT front-end to all of the WebSphere Application Server PMI statistics that are relevant to messaging performance. When you start the tool and point it at a server and messaging engine, it selects a useful set of default statistics. You can then use the configuration menu to select additional statistics with an indication given of the level of performance impact of doing so. Best of all (at least for someone as forgetful as me) is that when you close the tool it turns off the PMI stats before shutting down.

My first WebSphere ESB cluster

Saturday, January 27th, 2007

The second instalment of the WebSphere Process Server and WebSphere ESB deployment patterns series is now available. This describes in considerable detail the steps required to configure a simple Process Server cluster. By removing the parts that are plainly not applicable, you are left with a set of good instructions for creating a simple WebSphere ESB cluster. You should, however, refer back to the first article to review when using this simple topology is valid. In particular, you should note that co-locating the SCA modules and messaging engines is generally only possible when you are not using asynchronous SCA i.e. the module imports are not using JMS. If you do use asynchronous SCA then the partitioning of destinations on the SCA.SYSTEM bus that occurs as a result of clustering the messaging engines can become a problem, with responses no longer guaranteed to get back to a partition that is accessible by the instance of the module waiting for it.