Looking to flesh out its open source SOA platform, JBoss Inc. has purchased the Arjuna Transaction Service Suite.

The Arjuna Technologies Ltd. transaction engine, which has a 20-year pedigree and has been updated to enable Web services transactions, will become a free and open source product and be branded as JBoss Transaction 4.2, available in the first quarter of 2006.

"All J2EE application servers require some level of transaction management," said Shaun Connolly, JBoss vice president of product management. "This enables us to tackle the kinds of detailed transactions that Web services haven't yet been able to get at. This is proven stuff. It's a bet-your-business technology."

The Arjuna engine - now branded as JBoss Transactions - comes complete with support for the proposed Web Services Transaction (WS-TX) and Web Services Composite Application Framework. It also has demonstrated interoperability with Microsoft and IBM products.

That's a really smart move - Arjuna's technology is very solid. So solid in fact that it has survived many business events: university lab, start-up company, acquisition by Bluestone, acquisition of Bluestone by HP, HP pulling out of the app server market, Arjuna spun out of HP again, and now ArjunaTS acquired by JBoss... I may be missing some steps.

I'm only wondering JBoss didn't acquire the whole company. Were they not interested in a good JMS implementation too ? :-)

That's a really smart move - Arjuna's technology is very solid.

Yes, this is huge not only for JBoss, but the Java middleware open source community in general as a *REAL*, production ready, first-class, and mature transaction manager is now open source and usable by the community.

Bill

Yes, this is huge not only for JBoss, but the Java middleware open source community in general as a *REAL*, production ready, first-class, and mature transaction manager is now open source and usable by the community. Bill

What about a *REAL*, production ready, first-class and mature JMS implementation then ? You're not interested ? It's gonna take a few more years to write it all by yourself. I would be interested in an open source version of that too.

Another nice thing about ArjunaMS is that it contains an embedded version of their transaction manager. That works great with Spring... oups, sorry, that's a dirty word for you. Seam. It works great with Seam.

Yes, this is huge not only for JBoss, but the Java middleware open source community in general as a *REAL*, production ready, first-class, and mature transaction manager is now open source and usable by the community. Bill

What about a *REAL*, production ready, first-class and mature JMS implementation then ? You're not interested ? It's gonna take a few more years to write it all by yourself. I would be interested in an open source version of that too.Another nice thing about ArjunaMS is that it contains an embedded version of their transaction manager. That works great with Spring... oups, sorry, that's a dirty word for you. Seam. It works great with Seam.

I guess you missed Ovidiu's post on JBoss Messaging? He linked to it as well. Also, I beg to differ about our current JMS implementation. We have customers running millions of messages per day through JBoss MQ. It may not be as fast as some of the commerical implementations out there, but it is robust enough to build most production applications. JBoss Messaging aims to provide a completely redesigned stack and has been in development for more than a year now.

I guess you have missed our Embeddable EJB3 project as well? It is really misnamed as it it really an embeddable version of JBoss 5 and CVS HEAD. It contains non-Arjuna TM, JCA, JMS, JNDI, and a EJB3 container that can run in junit tests, Java SE apps, Tomcat standalone, and we even got it working with Weblogic. We aim to continue this embeddability with Arjuna TM. It is bootstrapped with JBoss Microcontainer, but it could easily be bootstrapped with Spring as well if somebody desires to put the effort in.

Just because we articulate some of the smaller projects of our JEMS stack like JBoss MC, EJB3, and SEAM as alternatives to a *vendor* like Spring, doesn't mean "Spring" is a dirty word at JBoss. In fact, we started a Spring/JBoss/EJB3 integration project and JBoss/Spring forum way back in August. Ales Justin, the maintainer, wrote a nice article about it and submitted it to TSS back in August as well, but TSS doesn't seem to put much priority on anything associated around JBoss that isn't controversial.

Bill

I guess you missed Ovidiu's post on JBoss Messaging? He linked to it as well. Also, I beg to differ about our current JMS implementation. We have customers running millions of messages per day through JBoss MQ. It may not be as fast as some of the commerical implementations out there, but it is robust enough to build most production applications. JBoss Messaging aims to provide a completely redesigned stack and has been in development for more than a year now.

I did not miss that post but as you say, it's still a while away before the initial 1.0 release, then it should take a year or two to mature. That's why I was wondering why JBoss had not acquired and open-sourced ArjunaMS as well. Of course I can understand that you feel the new JBoss Messaging implementation is too far along to change course.

I guess you have missed our Embeddable EJB3 project as well? It is really misnamed as it it really an embeddable version of JBoss 5 and CVS HEAD. It contains non-Arjuna TM, JCA, JMS, JNDI, and a EJB3 container that can run in junit tests, Java SE apps, Tomcat standalone, and we even got it working with Weblogic. We aim to continue this embeddability with Arjuna TM. It is bootstrapped with JBoss Microcontainer, but it could easily be bootstrapped with Spring as well if somebody desires to put the effort in.Just because we articulate some of the smaller projects of our JEMS stack like JBoss MC, EJB3, and SEAM as alternatives to a *vendor* like Spring, doesn't mean "Spring" is a dirty word at JBoss.

I know, I was joking about the animosity that has been exhibited formerly by various people on both sides.

In fact, we started a Spring/JBoss/EJB3 integration project and JBoss/Spring forum way back in August. Ales Justin, the maintainer, wrote a nice article about it and submitted it to TSS back in August as well, but TSS doesn't seem to put much priority on anything associated around JBoss that isn't controversial.Bill

Or, for that matter, anything that isn't controversial. You have to find the right title for the announcement, or somehow link it to AJAX crap. Example: "JBoss / Spring integration: is it gonna kill IBM's middleware business ?" or the non-sensical "Spring integration to bring revolutionary AJAX capabilities to JBoss!".

I'm only wondering JBoss didn't acquire the whole company. Were they not interested in a good JMS implementation too ? :-)

JBossMQ is currently being phased out and replaced with JBoss Messaging, which has has a good chance of becoming a very good JMS implementation. The estimated release date is Q1 2006 and the project web site is http://www.jboss.com/products/messaging

Ovidiu

This is good:

"As part of the deal, JBoss also brought in Arjuna's chief architect Mark Little, who will become the JBoss director of standards and head up the company's enterprise service bus project"

Does this use coordinator-cohort for availability?

Does this use coordinator-cohort for availability?

We helped pioneer some of the replication and transaction integration work and implemented a framework to support a range of different replication protocols (because one-size doesn't fit all). Check out http://www.cs.ncl.ac.uk/research/pubs/books/abstract.php?id=45 and http://www.cs.ncl.ac.uk/research/pubs/books/abstract.php?id=155 for instance.

It's a good move for Jboss, and should completely eliminate a known weakness (transaction recoverability) in Jboss server.

Peace,

Cameron Purdy
Tangosol Coherence: Clustered Shared Memory for Java

This is a great move for JBoss and the OS community.

The weird thing is that it took almost 2 days to show up here...

The weird thing is that it took almost 2 days to show up here...

No, it is not weird to take two days to show up here on TSS.

Even "http://www.theserverside.com/news/thread.tss?thread_id=23585" in January 2004 took two days (between Jan 26 and Jan 28) to show up here on TSS. You didn't say nothing to complain at that time, Jason?? You should not say anything now.

Jason, will you be willing to post some outside news here on TSS for us? Say yes, then you must clock them very aptly.

The weird thing is that it took almost 2 days to show up here... No way.

This is a great move for JBoss and the OS community. The weird thing is that it took almost 2 days to show up here...

Jason, why didn't you post it? Paul __ did, which is appreciated, but we had a longer piece in the works that we chose to go with instead.

Wonder how JOTM or Atomikos will react.

Well, I have just added JOTM to Tomcat 5.5 and it was quite a struggle. Terrible documentation, terrible examples, and a bug that hangs Tomcat shutdown that's been known for many months.

It will be interesting to see which of the following happens:

1) This will motivate the JOTM developers to do something about the lack of quality documentation and WORKING, modern examples and their product will survive,

OR

2) The JOTM developers will breathe a BIG breath of relief now that a very nice TM is available for free and they simple let the product die a slow death when everyone moves to JBoss Transactions.

Which side of the bet is your money on?

CaC

Ooook, but this was not the question.

The question was, how does it compare to JOTM and Atomikos ?

To answer it, you need some experience using JBoss Transactions (or even Arjuna). So what ?

Anyone around with some field experience ?

Is it a real good tool (that can be used in standalone mode as well, I mean outside of a J2EE server), or merely JBoss marketing again ?

Ooook, but this was not the question.The question was, how does it compare to JOTM and Atomikos ?To answer it, you need some experience using JBoss Transactions (or even Arjuna). So what ?Anyone around with some field experience ?Is it a real good tool (that can be used in standalone mode as well, I mean outside of a J2EE server), or merely JBoss marketing again ?

ArjunaTS will continue to be a stand-alone transaction service. That's how it started out in life (way before J2EE was even a glint in Sun's eyes) and we've managed to keep it that way so far. Hopefully this move will encourage other people to check it out and use it, even if they don't need all of JBoss AS.

Well, I have just added JOTM to Tomcat 5.5 and it was quite a struggle. Terrible documentation, terrible examples, and a bug that hangs Tomcat shutdown that's been known for many months.It will be interesting to see which of the following happens:1) This will motivate the JOTM developers to do something about the lack of quality documentation and WORKING, modern examples and their product will survive,OR2) The JOTM developers will breathe a BIG breath of relief now that a very nice TM is available for free and they simple let the product die a slow death when everyone moves to JBoss Transactions.Which side of the bet is your money on?CaC

I like to think we've got some good documentation. Once it's up on the JBoss site, check it out.

I like to think we've got some good documentation. Once it's up on the JBoss site, check it out.

Mark,

Congratulations once more on the Arjuna deal with JBoss.

Concerning the previous posts in this thread: I think our Atomikos documentation isn't that bad either (not to mention the out-of-the-box installation and startup we like to think we have), and we've had recovery as from the very first release 5 years ago -- that recovery also encompassing compensating transaction mechanisms from the very start:-)

Granted, we don't support JTS (actually we did have this for a short while but none of our customers asked for it -- so we dropped it).

As for WS-CAF: I was a fan at first (taking a modest part in the OASIS meetings whenever I could -- you know that Mark;-), but since none of the big (traditional) transaction players back it I don't think it stands a chance against WS-T. But if CAF makes it through, we'll have no problem in supporting it.

In any case, I am looking forward to continue having a good and sound relationship in the future - be it with Arjuna or JBoss.

Cheers,
Guy

I like to think we've got some good documentation. Once it's up on the JBoss site, check it out.

Mark, Congratulations once more on the Arjuna deal with JBoss.Concerning the previous posts in this thread: I think our Atomikos documentation isn't that bad either (not to mention the out-of-the-box installation and startup we like to think we have),

Hi Guy. Thanks for the congratulations.

I'm sure you Atomikos docs are good. I'm pretty sure I didn't say otherwise in my earlier post though.

and we've had recovery as from the very first release 5 years ago -- that recovery also encompassing compensating transaction mechanisms from the very start:-)

As you know, our ArjunaCore component within ATS supports loosening of the various ACID properties, c.f., Gray's matrix in his original IBM research report. That's how it's been at the core of all of our transaction products, whether they be JMS (AMS), OTS/JTS, JTA, or the various flavours of Web Services transactions (BTP, WS-TX [both WS-AT and WS-BA], and WS-CAF [ditto])

Granted, we don't support JTS (actually we did have this for a short while but none of our customers asked for it -- so we dropped it). As for WS-CAF: I was a fan at first (taking a modest part in the OASIS meetings whenever I could -- you know that Mark;-), but since none of the big (traditional) transaction players back it I don't think it stands a chance against WS-T. But if CAF makes it through, we'll have no problem in supporting it.

I think WS-CAF achieved one of its objectives: we're now into OASIS with WS-TX :-) And I really think Web Services people should take a look at WS-Context!

In any case, I am looking forward to continue having a good and sound relationship in the future - be it with Arjuna or JBoss.Cheers,Guy

Ditto.

Wonder how JOTM or Atomikos will react.

Frankly, from a competitive point of view I don't see any real difference with the situation as it was:

-ArjunaTS, previously owned by Arjuna, is now owned by JBoss
-high-end customers will still pay for it as they did before (because they need support and JBoss -- after all -- is a company that needs revenue like any other)

Granted, ArjunaTS will now be open sourced. But people who need transactions don't care what it costs - as long as it works. The whole point of using transactions is to avoid having to hack around yourself... At least that is what the JBoss marketing page says;-)

Best,
Guy

So, how does it compare to JOTM or Atomikos?

So, how does it compare to JOTM or Atomikos?

If anyone's looking for flame-war material, then hopefully this post won't be the one!

I've looked at JOTM several times over the years and monitored its evolution via their mailing lists. I've also been involved in some EU projects with the ObjectWeb guys. And I've known Guy Pardon (Atomikos) for several years through various interactions whilst I was distinguished engineer at Hewlett-Packard and later when we were starting up OASIS WS-CAF.

I think that both JOTM and Atomikos have matured over the past few years and, now that JOTM has a recovery system, they can both be classed as transaction managers: if you don't have recovery, then there's really no point in using transactions.

As for ArjunaTS (ATS), well check out arjuna.ncl.ac.uk->publications and you'll see that we've been developing it for quite a number of years. It's always had recovery and we've always been pushing the standards, whether that's in the OMG, JCP, OASIS, W3C or GGF. The system has been put through its paces many times over the years, particularly when it was acquired by Bluestone and later HP.

For legal reasons I can't go into specifics, but while in HP it brought in the odd few dollars from some pretty big customers. Once we span out of HP, I think Arjuna Technologies has a good customer list too, particularly for a small start-up.

Technically ATS is JTA/JTS compliant, with all the usual bells-and-whistles. It supports Web Service transactions - we did an interoperability workshop with the likes of IBM and Microsoft at the start of this year too. Plus, because it's been around a long time and we've had a fair amount of customer feedback, there's a lot of sundry material, such as good documentation, trailmaps, demonstrators, and thousands of QA tests.

I hope this answers your question.