Thursday, April 10, 2008

Web Services, EJB and Batch in WebSphere 6.1

As JAX-WS is a feature pack for WAS6.1, we first discussed the importance of feature packs for WAS.
Apparently, a feature pack is an enabler for innovation while not disturbing platform stability. IBM is committed to support the feature packs as long as the main version is supported.
Feature packs tend to find themselves into a future version of the product.
JAX-WS feature pack has till now 15k downloads!
This current f.p. is based on Axis 2.0 and brings a new standard to replace JAX-RPC old standard for WS interaction programming model.
More over, JAX-WS 2.1 that should be out soon, will be integrated into JSE 6.0 - this means that pure java modules will be able to consume web services without the need to add software libraries (did anyone say that dot.net does that for years already?).

EJB 3.0 is another feature pack for WAS6.1. It brings a much simpler programming model and allows concentrating on the business logic and not the container meta data.
Note that the two feature packs do not integrate well - an EJB 3.0 bean can't expose services using annotations ...

One interesting issue in EJB3.0 is the new persistence framework, JPA. It is based on the implementation of Apache OpenJPA, developed together with BEA, based on the core of Kodo product, acquired by BEA years ago.
IBM & BEA cooperate on this Apache project.

Batch processing has a focus for version 7 of WebSphere. Seems like the business case for Batch is quite clear. Also, the batch sequence of actions seems to be much the same in all verticals. This lead to a bean-based framework, rather simple batch programming model.

5 comments:

Chris Vignola said...

No need to wait for WAS version 7 if you're looking for java batch processing. IBM already offers a comprehensive java batch processing solution for WebSphere v6.1. It's called WebSphere XD Compute Grid. This product adds a full-featured batch processing infrastructure to your WebSphere environment. You can find out more about WXD Compute Grid here:

http://www-128.ibm.com/developerworks/websphere/techjournal/0804_antani/0804_antani.html

and here

http://www.ibm.com/developerworks/websphere/techjournal/0801_vignola/0801_vignola.html

Happy "batching" !

Chris Vignola said...

Doh! Those links got truncated in the post. Here they are in full:

http://www-128.ibm.com/developerworks/
websphere/techjournal/0804_antani/
0804_antani.html

http://www.ibm.com/developerworks/
websphere/techjournal/0801_vignola/
0801_vignola.html

Unknown said...

Chris,
Thank you for the insightful and correct remark.
WXD is indeed an important part of the grid/utility computing solution for WebSphere based resources.
It is a powerful product that allows also to manage non-WebSphere resources.
Main conceptual difference is the topology / licensing in which WXD requires other machines and software licenses.

Anonymous said...

JPA the EJB 3.0 persistence framework is based on Hibernate, to my knowledge. Why do you say that it is based on Kodo?

Unknown said...

Hibernate most probably contributed the theme and urgency to deal with the persistence layer of EJB. Yet, as it seems (and I quote others) the committee took a contribution made by BEA of its acquired O/R tool KoDo. Based on this product they've defined the base grounds for JPA and ofcourse each vendor then took his own way to implement. BEA placed Kodo there, JBoss placed Hibernate and IBM, as far as I know, developed all from scratch.