We spoke about it in our reviews of the papers (those of June 11th and that of June 18th), debate is very enlivened at present around a hypothetical standardization of the pile of standards SOA.

The occasion to fall again on an article of Aur lien Pelletier:    . This ticket in more than two years but question is more than ever of actuality at a time when more and more voices rise around the alternative solutions, in EJB on one hand, in Web Services on the other hand. It is worth being asked, enlarged and delayed on perspective: Are standards Web Services in SOA what EJB belonged to J2EE?

So put down, question brings several axles of reflexion:

Standards Web Services and EJB: the same reefs?

The points of comparison raised by Aur lien Pelletier are very pertinent and show that Web Services and EJB share a sad history:

• Too ambitious and too complex specifications.
  Answer thoroughly and there only once such broad problems; ambition is noble, but isn't it to fish by pride? Why not to move step by step by working by short iterations in a resolutely agile step?
  The nicest example of the foibles of this ambition holds in the existence of WS-I.org (). Is this not a roof space what an organism was taken up to make interoperable the different realisation of a standard of interop rabilit ?

• The promise of simplifying tools.
  Whether it is for EJB or Web Services, the beginning was more than difficult. The tools were there but did not hold his promises. Who has not ever had problem with a descripteur of deployment EJB 1 generated by IDE? Why option  to Generate interoperable WSDL  was not checked, for a long time, by default?

• A too heavy machinery for the majority of the needs.
  A lot of plans using EJB and/or Web Services use only a very small part of the capacities of these technologies. Simply because they do not need it. They join that the syndrome here of  the pecker to pick daisies  who encourages to use WAS (WebSphere Application Server) or WLS (WEBLOGIC Server) where Tomcat would be enough amply.

Why to speak about EJB to past?

The death of EJB, a debate which does not lack pepper! This subject is part of inevitable debates which ignite our community regularly and more modestly exchanges between consultants Xebia. Here are some elements of reflexion:

• Firstly, does EJB exist?
  Yes, there is only to look at their sulphurous reputation to persuade himself of it.

• EJB have 3 a future?
  Nothing is less sure. SpringFramework is a very serious rival already with several years beforehand and distance becomes hollow every day. Besides, unlike Hibernate who plays the card JPA, SpringFramework comes as an alternative (  WE STILL BELIEVE EJB DOES NOT ADD worth  By Rod Johnson IN Spring Pitchfork FAQ). They will note however, , that Interface21 joined JSR 316 (Popular dance EE 6).

• Finally, who supports EJB 3?
  Apart from JBOSS and Sun, there is not high society ready to fight beak and nail for them.
  For his/her part GAPED support EJB 3 since the exit of WebLogic Server 10, already some months ago. At the same time, to IBM, ONE does not wait before the first quarter 2008 for EJB 3 under WebSphere.
  Anyway, the editors turn currently more attention to SCA (Component Architecture service), SDO (Objects Data service) and other one models in the course of standardization.

In the meantime nothing was invented of better and of simpler and more competitive to spread a transactional context remote and a context of remote security.

We will come back more in detail in a next ticket on the future from EJB

What future for WS-*?

The fashion of any Web Service fast headed with lack of the triptych of departure (WSDL, SOAP, UDDI) which is not more than a diptych (even if UDDI is supposed to get up again since version 3). For example the transactional aspects are away there what imposes to manage mechanisms of compensation.

It is while WS-* started to flourish. Even after strengthening and purification, the confusion around these standards is palpable. It only has for persuading itself of it to read articles  Making Sense of all these Crazy Web Service Standards  published by Michele Leroux Bustamante on InfoQ or    published by Steve Craggs on the blog of Lustratus Research.

WS-* should not disappear of so early, but it would have to be for good reasons and not only because software industry invested there in great numbers. Their promoters will have to take up following challenges:

• Finalise specifications: A lot WS-* are still far from a version 1.0.
• Prune verbose or useless specifications.
• Give interoperable realisation.

Anyhow, adoption by all of an unique layer of standards Web Services remains a phantasm as explains it Rich Seeley in its article    .

To finish

Let us plant a not so opened door as it: Which is interest to display its services on an interoperable standard (or almost):

• As part of S.I. (or of one under domain of S.I.) mono-technological?
• When these services do not have vocation to be invoked by partners?