HP Exits J2EE Application Server and Web Services Markets

Discussions

News: HP Exits J2EE Application Server and Web Services Markets

  1. After a two-year effort, HP is shutting down its line of NetAction products which included the Bluestone J2EE application server it had acquired in October 2000 for $470 million. According to Giga, HP had 4% of the appserver market. HP will instead focus its J2EE efforts on its OpenView network management tools.

    Also being dumped is the Netaction Web Services Platform and the Web Services Registry.

    Read HP Shuts Software Unit Bought for $470 Million.

    Read HP exits e-business software market.

    Threaded Messages (53)

  2. Does HP's move mean anything to the J2EE industry? E.g. what does this mean to the business model of so-called 'free-as-in-french-fries' application servers? It's a shame to see the Bluestone products (which were, to say the least, quite innovative) come to such a sad and sorry end ...
  3. Survival of the fittest at its best... and continued consolidation of this sector. As long as there is sufficient choice and the remaining app server platform vendors continue to compete and innovate, this will actually help foster new and emerging companies to create innovative value-add solutions on top of robust J2EE infrastructure.
  4. <
    I certainly did not think that they were very innovative at all. To create a web application, you had to create an ServletEngine class (why??), which was also required in the url (something you may want to hide).

    sajava.ini and apserver.txt always felt awkward and clumsy like an old windows 3.1 ini file

    we had no end of trouble getting a jakarta-struts based web application up and running in a solaris machine (could not read resource files).

    We concluded that Total-e-Server stood for total-excrement-server and switched over to Jetty.

    I am not sorry to see Bluestone go at all.
  5. Bluestone has worked very well for us for several years. The J2EE application server marked collapsed along with the collapse of the dot com sector.

    Bluestone's lucky they got bought when they did. HP never seemed to know what to do with Bluestone once they got it.
  6. Can anyone remember the names of the freeware relational databases that competed in the early-mid 90s' against Sybase, Ingres, Oracle, and Informix?
  7. This is not a surprise to me at all :)

    http://www.theserverside.com/home/thread.jsp?thread_id=1662#5108

    The world is only big enough for 3 enterprise RDBMS, and by the same token can only allow BEA, IBM, and .NET (sorry guys) app servers. Plus the hardware vendors never have a clue about how to market software (IBM excluded).
  8. The world is only big enough for 3 enterprise RDBMS, and

    >>by the same token can only allow BEA, IBM, and .NET
    >>(sorry guys) app servers. Plus the hardware vendors never
    >>have a clue about how to market software (IBM excluded).

    So true ! A mature market is one where a small number of competitors exist and we have reached that point in the app server and database market. Regarding hardware manufacturers knowing what to do with software witness Sun and Java. No doubt the 'open' and 'almost free' aspects of java have accelerated its acceptance in IT but boy don't you think they regret not making more $$$ out if it like M$ have with their products ? Not that I'm complaining, just making an observation. Will be interesting to see if Sun keep with the iPlanet/Sun ONE app server or just throw in with BEA like HP have. When you change the name of a product you know you have problems.....
  9. Don't believe them when they tell you that Carly Fiorina, the towering dynamo of corporate leadership, secretly plans to acquire Presto's Fry Daddy unit and dominate the market in Java-enabled deep fryers.
  10. Hey, don't forget the open-source app servers like JBoss in your list. We'll be here long after IBM and BEA drop their app server offerings too. :)

  11. When IBM and BEA drop their app servers, J2EE (and maybe Java) would have probably died.

    If JBoss still exists then, it will be just the name around another product. But with Java dead, it would be some other Boss. (NBoss) ;-)

    J2EE needs IBM and BEA to survive, weather you like it or not.
  12. J2EE is here to stay and will continue to gain in strength. Its probably the .NET plaform which will die its natural death in its infancy. I will go to the extent of calling it "still born". So no amount of wishful thinking on part .NETers is going to cause any damage to J2EE
  13. <J2EE needs IBM and BEA to survive,>

    I think you've got it wrong. J2EE needs JBoss to survive. IBM and BEA will be the death of J2EE with their high license fees. Take a listen to the interview with JBoss leader, Marc Fleury, on this web site.
  14. I guess I'll get some flames for this, however...
    First, I think JBoss is a nice AppServer for small-scale projects. I don't think J2EE is only suited for really large projects, it can also save a lot of time and add quality to smaller projects. This is where JBoss is currently great. To be used for larger applications it still has a long way to go, and I really didn't like their 3.0 release policy (they knew there were quite some bugs). Also sorry to say, normally I like motivated people who believe in what they created... but Mark Fleury is a zealot and IMO not good for JBoss' image (although he is a technical genious, no question). But, currently JBoss is the hanger-on and WebLogic, WebSphere, Oracle IAS and maybe iPlanet are the upholders of J2EE. This may change quite soon though (bugfixing for JBoss, J2EE certification, good ECPerf results, admin tools (especially for clusters)), but JBoss definitely suffers from what happens with Linux, synonym for open source (problems in the community, unstable releases, and the "just another Unix" as e.g. IBM begins to distribute it; it begins to lose the "guru-charme" and people discover that OpenSource software is also "just software": It has bugs, is sometimes difficult to use, not free (cash) at all (support), ...; short: OS software does not only have advantages and neither is it "perfect" per se)
    Further (an example), we really tried to use JBoss for a mid-scale project but then encountered a problem and no one could solve it... they switched to BES then (please don't ask me which problem it was, I wasn't involved that much into the project).
    And for the smaller projects (because of the current situation within the IT market USD 10K/CPU is really too much... and this is the _non-clustered_ version!!) I liked HPAS really better (don't tell me Bluestone TES is crap, I'm talking about HPAS 8.0, and the name change did not indicate any problems, just the opposite) than JBoss... it is faster, scales better and easier to use. I hope it will somehow live further, it is a (technically) great AppServer, if it gets some more time to mature (with release 9.0, I have some insight here) it could even be the best AppServer (I doubt this will happen after this announcement though).
    A small comment about the DBs: I really don't understand anyone using MySQL or PostgreSQL. Both are good products (although I would put MySQL into the "small" category, although I know quite some large sites are powered by it), but IMO SAPDB is _by far_ better... and also OpenSource. Try it, too few people know about it!
    So, all together I'd really be disappointed if HPAS 9.0 was never released :-( IMO it was _the_ free AppServer, free and feature-wise (core J2EE features, not vendor-stack: Although they had great plans and even cool betas it was never finished) close to IBM and BEA.
    What I really don't understand is: What the hell did HP expect? That they'd write a new AppServer in one year and it would be profitable immediately? Or even No. 1??? Of course such a big software product needs some time, with 9.0 (maybe 9.1) this would have been the case. They could have really gained market share I think (a great AppServer for freemĀ“, a big company behind it with lots of Hardware sales and I think OpenView could have pushed HPAS quite a bit). Oracle did make it, why shouldn't HP have a chance? They should just have given it a bit more time, but what they did was really the worst thing imaginable; Invest millions of dollars for the aquisition of BlueStone, invest again to rewrite it, invest and develop a big software stack on top of it and now that they are close to being there they stop it!!! (Both 9.0 and the stack were close to be finished!).
    They also didn't make it easy for both customers and partners with their always changing strategy...
    Okay, thats it, sorry for the long post and thanks for reading ;-) but I'm really a bit frustrated that, lets say one of the top 5 AppServers (and, see above, maybe even the best) and a great software stack just vanish short before they are mature enough.
    Any thoughts?

    cu,

    Messi
  15. This is not meant to be a flame, but really - unless you can provide some facts like what tests did you run to show that HPAS is fater than JBoss, or what specific problem did you have that JBoss could not solve and BES could solve - your post is so much blah, blah, blah.
  16. As I said, I don't have enough insight into the project to point out the specific problem (but I trust my fellow developers that there was a problem when they said there was one).
    In general, with JBoss they often had a hard time configuring it correctly, because although the config files are well structured they are still large and documentation of JBoss is lacking (but AFAIK this got better with 3.0).
    Regarding the speed HPAS vs. JBoss: We wrote a mid-sized application for a customer, something ERP like, and serached for a cheap software solution and first thought of JBoss (2.4) with Interbase but then tried HPAS with Interbase (finally we chose HPAS with SAPDB) and it was quite a bit faster than JBoss... although this may have been specific to the application I don't think so as the structure of it was quite "average". Using SAPDB made it even a bit faster.
    Please understand that I cannot release the source code ;-)

    regards,

    Messi
  17. <Please understand that I cannot release the source code>

    And please understand that your comments are not credible for the reasons mentions in my previous post.
  18. <quote>
    And please understand that your comments are not credible for the reasons mentions in my previous post.
    </quote>

    Greg -
    Those comments seemed to me perfectly credible - they were one person's experience. If you have had good experiences with JBoss on larger projects, then maybe talk about those instead.

    Cheers
    Ray
  19. <Those comments seemed to me perfectly credible>

    I think you should learn what credible means. They are not credible. They are possible, but thats not credible. They are understandable, but thats not credible. Credible means what are the reasons I have to believe him. I don't have any reasons to disbelieve him, but that mere fact is not a reason to believe him. Just because a person makes a statement that server A is faster than server B does not make it true. I want the specs. Perhaps I want to duplicate the test. If I based my decision on what server to choose based on his "credible" statements, I'd be drummed out of the business. Just trying to reduce the hype. I want something I can base a decision on. And if I can't base my decision on his statements, what good are they?
  20. Hmm, I didn't ask you to base business decisions on my statements, but still, I think they are credible (what would it buy me lying?)... You can easily imagine that I'd really like to give you something more "credible", but still... please understand that I cannot give you the source (and I think you _do_ understand that).
    I just wanted to share my experience in this area (yeah, I also wanted to comment, correct), and BTW: Its really not that difficult for an AppServer to be faster than JBoss (see Csiro report, this one is credible for you I hope, they give you lots of facts).
    All I can recommend to you is benchmark both servers with _your_ application, and unless you have done this I think "HPAS is faster" is more probable than "JBoss is faster" (though, as I said, it could be only true for our specific application).
    In fact, I'd really _encourage_ you to test both servers and tell us what you found out (and, if possible, release the source for the benchmark)... I'd really like to know how these two servers perform with another application.

    kind regards,

    Messi
  21. Look, this is really simple logic. For me to say that that its not credible in no way implies that I think he is lying. Sheesh. And why should I run some tests to prove that you are wrong when I don't believe that you are wrong ???? I just want you to prove to me that you are right !! Simple Logic 101. For you to report your experiences without proof may be interesting, but is not useful in the least. I'm outta here.
  22. Greg -
    Thanks for the vocabulary lesson!

    He's just making statements about his experiences. If he experienced it - it is credible to me that he is telling the truth, since credible means offering reasonable grounds for being believed. His group didn't have success with JBoss - what's so difficult to believe about that? I also know people who have had success with it - that's not difficult to believe either. Unlike yourself, I guess I wasn't trying to make a business decision on his statements.

    Cheers
    Ray

  23. <Unlike yourself, I guess I wasn't trying to make a business decision on his statements. >

    If I can't take anything away from the comments to help me make a decision on what app server to use, what use are his comments ?? As I said before, it may be entertaining, may be interesting, but in reality, its so much blah, blah, blah. And there is far too much of that out there today, people making decisions based on blah, blah, blah.

    Just found this book at the bookstore and bought it. Looks pretty good. Reminded me of this discussion:

    "J2EE Performance Testing"
  24. <If he experienced it - it is credible to me that he is telling the truth, since credible means offering reasonable grounds for being believed.>

    Your definition of credible is a good one. But the fact that he experienced it is, I don't think, reasonable grounds. I mean, all you gotta do is ask a several people, all of whom experienced the same event, to describe what they experienced and you will get all different experiences.
    Lets hope you are never on trial for murder, and the witness is up there on the stand saying he witnessed you doing the evil deed and your lawyer says, "Well if he witnessed it, it must be true. I won't ask him anything. Like, oh you wear glasses but on the night in question you wern't wearing them? Oh, the oncoming traffic headlights was in your eyes?"

    Again, not saying that people are lying, they aren't. But experiences are very, very subjective.
  25. This whole thread is in-credible.
  26. <quote>
    A small comment about the DBs: I really don't understand anyone using MySQL or PostgreSQL. Both are good products (although I would put MySQL into the "small" category, although I know quite some large sites are powered by it), but IMO SAPDB is _by far_ better... and also OpenSource. Try it, too few people know about it!
    </quote>

    Absoultely 100% _Agreed_ on SAPDB. Grab it at:

    http://www.sapdb.org

    A list of some of the features is at:

    http://www.sapdb.org/sap_db_features.htm

    Cheers
    Ray
  27. The one thing I like the most about SAPDB:
    About 1000 R/3-installations use it world-wide. These guys pay license fees for their R/3-license and maintenance fees for SAPDB (SAPDB costs maintenance only when used together with R/3).

    What does that mean? Well, it means that SAP is committed to these customers in the long term and will maintain SAPDB for a very long time. So, it is the GPL RDBMS with one of the longest expected lifespans.

    Cheers,
        Henrik
        TNGtech
  28. And not to forget that SAP DB has a long history. In the early 80s, when I was a young and smart developer, I worked with it at Nixdorf Computer. It was developed at SQL Datensysteme GmbH in Berlin, a subsidiary of Nixdorf, and the name was DDB/4. We've used it as a pilot version under VM/CMS that time and it was already very good. Then it was going on a long journey after it finally labelled SAP DB.

    Check out the History

    -- Andreas, still smart
  29. I don't know of any statistics or hard facts that support your statements about HP "scaling better" than whatever other server you were abstractly comparing it to.

    And why did HP stop the BlueStone thing? Because even though I am sure that HP's J2EE product was pretty good, the J2EE industry *needs to consolidate*... obviously, since HP has chosen BEA to be their "App-server" partner, HP believes that WebLogic is a good substitute for BlueStone.

    It would do this thread nicely to find out exactly why HP chose BEA to partner with rather than going to other J2EE vendors.

    Basil.

  30. <B>OS <-> App Server</B>

    AIX <-> WebSphere

    Solaris <-> Sun One

    Windows <-> .NET

    ??? <-> WebLogic

    HP UNIX <-> BlueStone (dropped, low market share)

    <B> New Solution </B>
    HP UNIX <-> WebLogic = competition


    Oracle will use their DB as leverage rather than OS


  31. See my above message, we tried it with this application.
    Don't know what you mean by "And why did HP stop the BlueStone thing?" BlueStone TES (because it was not very good) or HPAS?
    I understand why they stopped HPAS (not enough market share, too expensive to develop), but I really don't get their strategy: IMO they stopped it when they began to get a chance to gain market share (as I said, you won't gain significant market share with a 1.0 release, and it will need some time).
    Sure, the market needs to consolidate (I fully agree), and sure, BEA and IBM are the big ones, but HPs starting position was at least not worse than Oracles. And they seem to make it.
    Don't know exactly why HP chose BEA in the first place, but now they chose it because they already teamed with them before Bluestone. I also think they chose BEA because IBM wouldn't be a good choice, agree?

    regards,

    Messi

  32. >It would do this thread nicely to find out exactly why HP chose
    >BEA to partner with rather than going to other J2EE vendors.

    I would imagine it's because WebLogic is far and away the most common app server running in production on HP hardware, and has been for years, throughout all of HP's efforts to market their own app server. (No, I don't have a public source to cite for that, though one may exist in some industry-analyst report; you'll just have to believe it or not as you choose. :-) A fact that is easily confirmed in public information is that HP is a primary platform for WebLogic, meaning that it is certified on that platform on the initial release date for a new version of the server. I doubt that's true for either WebSphere or iPlanet.

    The choice seems pretty simple, really. They had tried both developing their own server and buying one with some market share and trying to build it up, and had not succeeded with either approach. (I'm making no judgment here about either the technical merits of the products or HP's handling of them.) If they didn't want to go down either of those paths again, starting even further behind, they had a choice of IBM or Sun, both of which would have a conflict of interest with HP's goal of selling servers, or BEA.

    The fact that WebLogic is a great product is just a bonus. :-)

        -- Jim
    (disclosure: former WebLogic developer)
  33. I don't understand nobody even talk about Oracle OC4J. It's really a great product, and with Toplink, I think it'll be the best one.
  34. JBoss, and Tomcat for that matter are like mySQL or PostgreSQL in the RDBMS space. They will be around as long as J2EE lives, and some people will use it for fun or even small-scale projects. But never expect large enterprises to use them for their "mission-critical" systems, because in people's mind $10K/cpu means high quality + lots of technical support whereas free means not reliable and no support (not my opinion but most CTO's tell you so - wouldn't you think the same way if you were in their positions?).
  35. <quote>
    (not my opinion but most CTO's tell you so - wouldn't you think the same way if you were in their positions?).
    </quote>

    Since I know that 10k/cpu doesn't mean high quality and that "lots of technical support" doesn't mean quality technical support, I probably wouldn't think that way. If a CTO makes a choice on those criteria they should be fired immediately since they aren't serving the needs of the company. It doesn't neccessarily mean that I would use JBoss for large mission critical systems, but cost, even in large systems, is much more of an issue now. I can see no real reason to use BEA or IBM in every large app context - some, maybe, but not all. If a CTO is going to help deliver a true cost effective value-added solution that is truly useful to the company, they should use their brains just a little more. Perhaps that is too much to ask.

    Of course, I doubt seriously that I could ever be a proper CTO ;-)

    Cheers
    Ray
  36. <>If a CTO makes a choice on those criteria they should be fired immediately

    Well, we better start making a list.

  37. <But never expect large enterprises to use them for their "mission-critical" systems, because in people's mind $10K/cpu means high quality + lots of technical support whereas free means not reliable and no support (not my opinion but most CTO's tell you so - wouldn't you think the same way if you were in their positions?).>

    No I would not think that way. With the economy the way it is, its forcing CTO to rethink their assumptions about $10K/cpu and thats a good thing. Furthermore, just around the corner will be ubiquitous small web aware hardware devices. If you had to ship 10,000 web aware devices, you gonna pay $10k/device - I don't think so. With the small footprint (70K) of the minimal JBoss super server with hot deploy and JMX admin - its a no brainer.
  38. What concerns me more is that someone wants to turn small web aware devices into things that need to be running ejb containers, is this somewhat reverse logic?
  39. <quote>
    With the economy the way it is, its forcing CTO to rethink their assumptions about $10K/cpu and thats a good thing. Furthermore, just around the corner will be ubiquitous small web aware hardware devices. If you had to ship 10,000 web aware devices, you gonna pay $10k/device - I don't think so.
    </quote>

    Precisely one of the scenarios Pramati is very well suited for- "Wide Deploy". With Pramati's aggressive ISV and Enterprise (volume) licensing, combined with the strong tech capabilities (proven by being the first J2EE 1.3 certified and only server to have non "data-corrupting" Ecperf numbers) this should be a good choice to consider in Wide-Deploy scenarios.

    Cheers,
    Ramesh
  40. Nothing will happen.

    It is good for hp and it helps them to concentrate in their core competencies.

    If a software project doesn't make money, it is better to shut it down.I am sure, it will not make any changes in the J2EE market.

    Lawrence
  41. I wish more companies would concentrate on their core competencies. :)
  42. "A small comment about the DBs: I really don't understand anyone using MySQL or PostgreSQL. Both are good products (although I would put MySQL into the "small" category, although I know quite some large sites are powered by it), but IMO SAPDB is _by far_ better... and also OpenSource. Try it, too few people know about it!"

    I agree with you on MySQL. But I like postgresql quite a bit. On top of that it has excellent support for geospatial datatypes. I can't get to sapdb.org right now. Does it support that?

    -Newt
  43. Not sure about the datatypes, I also like postgresql, especially lots more than mySQL. Still, I found SAPDB to be better for J2EE solutions (HPAS with SAPDB was a great combination).
    SAPDB also has better platform support than postgresql (it really only runs on Linux).

    regards,

    Messi

  44. spoken like a true pramati zealot :) Did you forget to say that you work for Pramati?

    seriously, I doubt pramati has the recognition to make it big in the US/Europe. They will remain a very small niche player at best.

  45. May be Pramati didn't get recognition in USA market but no doubt its the best app. server so far. and i am sure people who have used it will agree with me.
  46. "spoken like a true pramati zealot :) Did you forget to say that you work for Pramati?"

    I sure work for Pramati. And am a sure Pramati Zealot :-)

    "seriously, I doubt pramati has the recognition to make it big in the US/Europe. They will remain a very small niche player at best. "

    That small niche could very well be the wide-deploy and multiple-deploy (ISV) segments. That certainly is our primary target market.

    Cheers,
    Ramesh

  47. JBoss, and Tomcat for that matter are like mySQL or PostgreSQL in the RDBMS space.

    > But never expect large enterprises to use them for their "mission-critical" systems

    I do not think that applies to the databases that you mentioned. MySQL powers a surprising number of commercial web sites( Yahoo! Finance, MP3.com, Motorola, NASA, Silicon Graphics, and Texas Instruments - from their web site), and PostgreSQL powers the site you are looking at at the moment! (Click on about the site if you don't believe me.)

    'Nuff said.

    /david
  48. Interesting insight, if your facts are accurate. Although it is hard to believe somebody like Yahoo runs on mySQL.
  49. <quote>
    Interesting insight, if your facts are accurate. Although it is hard to believe somebody like Yahoo runs on mySQL.
    </quote>

    Check out this link:

    http://www.mysql.com/articles/us/yahoo_finance.html
  50. Hard to believe? MySQL is a rock solid lightening bolt.
  51. Very true, world is only big enough for 3 SOFTWARE (OS,DBMS,AppServer, Apps) companies M$, Oracle and IBM.

    Regarding (J2EE) Application server, sooner or later either Oracle or IBM will buy or kill BEA to get it out of the way.

    Sad, but true.
  52. Or may be it just spells a dooms day for entire J2EE community. I guess the Party is over for IT industry.
    Sorry if I am sounding too pessimistic.
  53. I hope they atleast open-source HPAS 8.0 and related technologies (like the CSF, and RadPak). I for one would like to see one more entrant in the Open Source AppServer Arena, JBoss could do with some competition atleast in the documentation, where HPAS 8.0 excels. (This is by no means an attempt to start another flame war...I love JBoss and use it for all development!) but it would be cool to modify HP's deploytool- the RadPak, to work with JBoss and add XDoclet integration to it (drool, drool).

    Sat Nashikkar
    JP Mobile
  54. Isn't the CSF already LGPL[ Go to top ]

    Last time I looked the CSF was LGPL.

    It's an interesting microkernel comparable to JBossMX. Furthermore, it's got a nice GUI for configuration.

    Carlos

    P.S. Isn't Bea a big time winner in all this, all of HPs appserver contracts are now going to go to them?