IBM focusing on a services-oriented architecture for Websphere

Discussions

News: IBM focusing on a services-oriented architecture for Websphere

  1. IBM plans to aggressively extend WebSphere AppServer from a vanilla J2EE to a services-based architecture that allows the building of apps with integrated workflow, business rules, and network-based transaction capabilities. Says an IBM official: "In the on-demand applications world, J2EE alone is like having a heart without the lungs. Putting a Web services veneer on top of J2EE is just not sufficient".

    Read IBM eyes services-based role for J2EE.

    Threaded Messages (36)

  2. J2EE future?[ Go to top ]

    This does not sound good to me.
    I see a J2EE future being threatened.
    With market consolidating, there will be only big brothers left.And don't expect them to be obedient in implementing specs from their competitor.
    As the applications get complex, developers ought to and will take advantage of frameworks offered by vendors on top of J2EE and that tosses out the utopian idea of WORA (write-once run anywhere), WODA (write-once run anywhere), blah blah...
  3. J2EE future is fine...[ Go to top ]

    I don't see this affecting the future of J2EE too much. If J2EE was supported by only a couple of vendors, then IBM's decision would probably shake the foundation a bit. Hopefully the market consolidation won't be so drastic that only a few players will be left.

    By moving forward, IBM might not follow open standards, but it will likely still contribute much to technology as a whole.

    Danger will come if they attract a very large customer base and decide that they really want to put an emphasis on vendor lock-in. Then J2EE would lose a big player.
  4. BEA's Platform Suite whereby workflow, process management, rule-based security and personalization provide productivity and managability on top of enterprise-strength J2EE and WebServices is a reality today for numerous BEA customers...

    Imitation, however, still is the sincerest form of flattery...
    Matt
  5. I don't have any idea how come somebody can compare IBM with BEA. IBM is a company with a lot of features that includes a great number of products, customers and handful of money. IBM can play a great role in technology and customer satisfaction of service oriented software business.
       In other hand BEA is a small player, other than their Weblogic application server, all products from BEA is nothing but collection of garbage. BEA usually tries to be over smart by using the market share they got during the .com time. People have to believe one thing, nowadays there are very small number of companies left in Bay area, which operate from one room office and dreaming about a web technology with investor’s money and making call to BEA office asking about one CPU license price.
       So boys and girls of serverside.com, let us stop comparing anything with a childish company like BEA. Let us talk something little more matured.
  6. I'm sorry but I forgot, weren’t you the disgruntled BEA employee I heard about?
  7. RE: to Disgruntled BEA employee[ Go to top ]

    Services-oriented architecture
        During the last 5 years there are a lot of changes happened in information technology, nowadays they are called service oriented architecture. As a software programmer/engineer/developer/architect what I wanted from a vendor?-
     I need a solution reasonable to the money I pay for the vendor including fully functional modules and services for all stages of software development such as
    · Business requirement Analysis
    · Business modeling
    · Design
    · Architecture of the system
    · Development
    · Integration
    · Deployment
    · Support and mentoring of end user
    If a company is not providing me all these requirements as a vendor, I don’t think they are vendors; they are just some group of people out there in market who try to cheat the customers and make money.

    Recent developments from IBM such as purchase of rational software, merging of one of the big five consulting firm, and outstanding integration with specific domain expert such as Siebel promising a good future of all solution under a single umbrella. In order to provide what customer needed from a vendor, they should have all integral components such as
      Design tools,
      IDE,
      Middleware,
      Database,
      Hardware and over all industry knowledge.
    IBM is in business of software for years and they still doing the major information technology business in hardware and software areas. I don’t have any idea why somebody has to compare IBM with BEA. BEA is nothing compared to IBM and IBM will not respond to all this childish comments in serverside.com. I have seen a number of BEA employees coming here and responding to the comments in an irrelevant way.

    Disgruntled BEA employee
      
    I have noticed some personal comments about me in serverside.com . I would like to respond finally to that.
    Those who use that words such as one mentioned above have to really go and study its meaning first, I am talking with some of my funny past experience such as
    1.I have traveled across California to a client place to add sql server database driver class jar file to class path (hope some big guys remember that sql server database driver classes were not part of weblogic.jar in earlier version) and save the image of a principal consultant from BEA.
    2.I have seen client say goodbye to principal consultant cuz he don’t know the problems in session tracking while accessing wlcs application using IP address.
    3.More over BEA used give me additional gifts for my outstanding work.
    4.I have came across some customer support people who told me, it is not easy to extend the commerce server database schema
    5.I have came across people who don’t know that a startup class is required to establish a gateway communication with CRM application in weblogic JAM.
    6.All these is very funny for me and I laugh when such a company is being compared with IBM.
    7.Finally, I have seen a white paper published recently on best J2EE practice, and an early drama for comparing per store application between .NET and J2EE. The white paper sounds like, they are still in J2ee classroom and pet store comparison sounds like a promotion or financial aid from Microsoft.
    8.Guys who make dirty comments on my views, remember, I have a successful client base, proven project records, a job even under worst economy. I don’t wanna comment anymore on these subjects and kids play your games in your backyards not in professional world.

    Thanks
    T.Q
  8. RE: to Disgruntled BEA employee[ Go to top ]

    Two words: Sour grapes.
  9. RE: to Disgruntled BEA employee[ Go to top ]

    If a vendor is telling me that they provide everything I need for every step of the development process, including analysis, design, development, and every stage of deployment, and that their solution will work for me, then I say that the vendor is lying.

    There is no "one-size-fits-all" solution out there that IBM or anyone can provide for enterprise development. It's just too complex. I think one of the advantages that J2EE has brought to the market is the ability to mix and match components from different vendors to find a solution that is most appropriate for a given situation. Given that, why would anyone go with everything from one vendor(unless they actually believed what the IBM/BEA/Oracle/etc/etc sales rep told them)? I thought that's why we worked in the J2EE world and left Microsoft behind. At least when MS tries to sell you a total solution, they don't make you pay $200/hour for consultants to help you install their products.
  10. RE: to Disgruntled BEA employee[ Go to top ]

    |
    | If a vendor is telling me that they provide everything I
    | need for every step of the development process, including
    | analysis, design, development, and every stage of
    | deployment, and that their solution will work for me, then
    | I say that the vendor is lying
    |

    Agreed. 100%.

    There is one thing that is a given. The vendor with the integrated stack will always be trying to sell you the whole stack. You can also be guaranteed that they have less interest in developing their products to easily integrate with competitors software.

    What is more, if the same vendor is providing consultancy services, there will always be some question over what vendor they will recommend. This is not necessarily due to any dishonesty - its just that its far more likely that the majority of their experience will be with their own products.

    Good products, that are flexible and easy to use, support the latest standards and that represent good value for money are all that most competant IT architects are looking for: A good tool that does a good job. Vendors selling "solutions" are selling a dream IMO.

    -Nick
  11. RE: to Disgruntled BEA employee[ Go to top ]

    <q>
    If a vendor is telling me that they provide everything I need for every step of the development process, including analysis, design, development, and every stage of deployment, and that their solution will work for me, then I say that the vendor is lying.
    </q>

    I could not agree with you more. This is not only true for vendors but applies anywhere. Noone can do everything well. IBM is trying to move from a software vendor to a service provider. Buying Rational may have been a good strategic move but is no reason to lock into IBM.

    I like to do my modeling with Together although I think XDE is a nice tool. This does not mean that I´d use JBuilder for development.
  12. |
    | Let us talk something little more matured
    |

    Yep. Good advice.
  13. I believe the reality is BEA only offers 'Simple Web Services'. It doesn't provide support for 'Business Web Services'.

    Providing 'Business Web Services', 'Autonomic computing' and 'Grid Computing' an Imitation? I do not understand...

    Best Regards,
    Murugan Kanpa
  14. Indeed, BEA Workshop started pushing forward some time ago. IBM states usage of WSIF but is being vague about when they will support BPEL which they seem to hint is the ultimate mechanism for implementing processes, workflow and alike.

    If BPEL can be implemented on top of a plain vanilla J2EE app server (see http://www.collaxa.com), why does IBM need to resort to making J2EE proprietary???

    Cheers.

    Jill.
  15. Matt,
    But Bea is missing a Rules Engine, Bea has a Workflow Product but not a Rules Engine (you can't count the one in Portal as a rules engine)

    Hopefully some pressure by IBM will force Bea to finish their product offering (with a rules engine)

    You are right that IBM is playing catchup, its just in 5.0 that they learned to spell JMX, a lesson everyone seems to have taken from JBoss...

    this sort of competition is good...

    -jp
  16. The BEA platform also includes a rules engine in what used to be the Commerce Server.
  17. Last year, at OMG Days, I heard a presentation from Patrick Maes, Chief Technology Officer of SoundView Technology Group. The study was about the economic view of the middlewares market.

    An interesting conclusions was that application servers are becoming the Application 2 Application integration layer, and that gives an advantages to the companies who have important market shares on the middleware market, and particularly on the connectors : 1st IBM w more than 30% of market share, 2nd TIBCO around 10%, all other players under 5%.

    Today, if you are interested in EAI products, you'll see that the offers turn around TIBCO and IBM with products like Crossworld & MQSeries. BEA is not in the 1st players.
    If you also look at corba, I don't think BEA has any ORB expertise in its offer. They also have no known offers on Databases, Security, Mainframe integration.

    So my question is, how can BEA be a big player, able to influence the J2EE specs, while most of the money still comes from the middleware market and the legacy integration.

    I worked for a french insurance corp, who wanted to integrate its mainframe legacy to its e-business. BEA sales offered very poor solutions (it was in 2001) compared to what other competitors could offer. They also were unable to give a price since their solution was still hard to integrate to their own application server.

    For the financial sector, BEA is not a big player. It is far behind Tibco (from Reuters), IBM, Microsoft or Borland.

    My second question is How can you sell a e-business or a business service oriented platform if you are not able to help your customer on legacy integration, while the customer's business logic is still in its legacy ?

    I also have to say that I never used BEA product, but their small catalog never gave me that opportunity.
  18. You couldn't be more right[ Go to top ]

    BEA offeres very little when it comes to integraqtion. Except for Tuxedo, they make it very hard to integrate with other legacy. (The majority being applications like CICS, IMS, VSAm, etc...) I have seen many customers have to use an extra layer of indirection (WLS JMS to MQ Bridge) just to get to the integration layer. They make it very hard to register external entities into their JNDI. (Although a good startup class with some JNDI coding can do it).

    BEA had a very weak CORBA product called M3 some years back that was well behind IONA and Visi-Broker. Hence being why they rely on the T3 protocol more than IIOP. They were smart to join Microsoft and IBM in the Web Service race.

    If you look at back in the Web Service days, IBM came uo with the service oriented architecure. (The SOAP, WSDL, and UDDI triangle). It was first announced on IBM alpha/developer works. So how can they be joining or catching up in an architecture the coined??
  19. You couldn't be more Wrong![ Go to top ]

    BEA's Integration offering kicks A**!

    Based on the Connector Architecture, JMS, WebServices, and XML technologies; It provides the most standards-based approach to integration by far.

    This standardization offers more choice, productivity, flexibility and investment protection than legacy proprietary solutions based on MQ or Tibco will ever offer.

    As far as mainframe integration goes there are numerous options available including J2EECA adapters, JDBC drivers, JMS bridges, CICS/IMS gateways, and even direct file access (Companies are able to run WebLogic Integration on LINUX on OS/390 and it rocks!)

    You say "IBM came up with the service oriented architecure". I say "so what", Xerox came up with the mouse and then what did they do with it?

    Matt
  20. no problem[ Go to top ]

    You can prefer BEA as it is more suitable to your needs. I don't think that one editor is the right choice for every body.

    I am just sceptical, when I hear you talking about BEA's offer, about the role that BEA can play in the IT technologies evolution.

    Your example of JCA adapters use for host integration is exactly the demonstration of what I am saying : the spec is originated from the Common Connector Framework from IBM.

    No one said that proprietary solutions is better. I said that the vendors who lead the adapters market are more able to influence and boost J2EE specs than others.

    BEA doesn't have significant levels of investment in research labs and open source projects, which is the real key to J2EE specs maturation.

    By the way, I am not working for any vendor. I used several distinct vendors products, from Sybase to Microsoft. I hope one day I'll have the opportunity to play with BEA.
  21. I think JBoss is at the heart of this movement towards a services oriented architecture of the future. Afterall, it was the first J2EE based container to utilize JMX at the core as a microkernel for cycling of "services". Each J2EE API is nothing more than a "service" or MBEAN, that plugs into the JBossMX core.

    This has been for over 1 year now. This vision is being extended in JBoss 4.0 in which you will be able to add EJB like behavior to any Plain Old Java Object (POJO) in your enterprise. Instead of having to define the EJB interfaces, all you need to do is give JBoss a POJO and we will do the rest. So we can help replicate state across clusters, add security, persistance, logging, transaction, etc to any POJO. This vision has been adopted by a lot of companies recently who see the value in this aspect oriented approach. By the way, this is not too different from .Net. This will also allow for plug and play enterprise services of the future.

    Granted, JBoss does not have the clout that IBM or Microsoft have, not even close. Or at least not yet:) So when we move in this direction, it takes the likes of IBM to back up this vision to create a standard.

    However, I would not blame JBoss or IBM for J2EE moving towards fragmentation. I would blame Sun for trying to be the owner of the standard and having a competing product for FREE at the same time. If I were IBM and BEA I would be doing the same thing, which we are. Every message you read out of Sun these days in my mind send a clear message, "WE ARE TRYING TO TURN OURSELVES INTO A SOFTWARE COMPANY" They think the future of Sun is Software.

    So if I am a company that makes software than competes with Sun who owns the standard and clearly wants to sell a product based on that standard,I would look to innovate beyond that standard. If Sun wants, which to their credit they haven't, they can slow the standards process down until they have their own products ready. This does happen all the time in the JCP from all the big vendors.

    We can continue to throw blame around here, blame .Net. Why .Net? Why not? Seriously, .Net is a QUALITY implementation of a distributed computing framework, from a company who has spent $1 billion to make sure they get it right. And has proven to make good development tools over a 10 year period. And they have a lot more where that came from. $43 billion and growing. You can only fight an implementation with an implementation not with a paper specification. Why has CORBA died? Too many people to agree on a common agenda when everyone has their own agenda. I admire Microsoft for being up front and letting everyone know they have their own agenda. I wish other vendors would do the same....

    In general, we have moved from a standards war to implementation war. J2EE has been very successful in creating a market and more importantly a framework for innovation in that market.

    With a statement like this from IBM and over 2,000,000 downloads of JBoss in 2002 alone, I think the future is VERY clear. It will be a service oriented future, both in software and in business. Quality Software and Quality Services will rule the future. IBM gets is, I think HP gets it, Microsoft and Sun seem to think that they can solve the worlds problems with software. Microsoft has enough cash to survive, does Sun?

    In general I hope the future of computing is far different from the past. The past consisted of a lot of false promises and investment gone bad. Hopefully the future consists of smart people solving complex problems on time and on budget. We need to realize that computing is just another services field using commodities and knowledge to do the job.

    So is J2EE dead? Not yet, but it I think it is not too far off. Unless Sun turns J2EE over to the community and steps aside. Let the Open Source community do what we did for the OS to the Enterprise OS.

    Cheers,
    Ben Sabrin
  22. very clear?[ Go to top ]

    <ben>
    With a statement like this from IBM and over 2,000,000 downloads of JBoss in 2002 alone, I think the future is VERY clear....
    So is J2EE dead? Not yet, but it I think it is not too far off. Unless Sun turns J2EE over to the community and steps aside.
    </ben>

    Funny that niether IBM nor JBOSS have a shipping SOA product even closely comparable to BEA's WebLogic Platform.

    J2EE is not going to die, but products that are stuck at ONLY implementing J2EE might -regardless of whether they are commercial OR open source.

    Matt
  23. very clear?[ Go to top ]

    Matt,

    You are correct, we do not have a platofrm because the industry could care less about the platform. If they did they would be waiting with baited breath for .Net.

    We talk to a lot of people who just want a quality app server. The difference is we don't sell software we sell knowldege. So, we could care less about pumping a platform to try and make license revenue. A lot of customers in the J2EE field are barely using EJBs and some very large ones at that. Yet, companies are trying to sell them on the idea of a platform that they do not need.

    I will agree with one point of yours, people spent doing just J2EE and nothing more will be left behind open source or not.

    The future will be fun, we are having a lot of fun.

    Ben
  24. very clear?[ Go to top ]

    I don't see where IBM is playing catch up in W when IBM and Microsoft were the first one defining the standards. Apache SOAP, WSDL, AXIS, WSIF, etc.. all came from IBM. IBM then released it to Open Source. We had the products and you could use them within WebSphere for a while.

    I admit as far as J2EE goes, WebLogic has been first to market, but saying that WebLogic is ahead is Web Services is a Joke, WSAD and WSAD Integration Edition has has Workflow and Microflow capability based on WSIF for over a Year. The WebLogic JWS standard thing is no where in the spec that I see (Just because it's a JSR doesn't mean it is going to be apporved). IBM was first with the Apache implementations. WLS's announced 8.1 release did not even tout J2EE 1.4. IBM has had Workflow Products for a very long time.

    As far as JBoss goes, JMX is a good technology but it is not the only thing out there. So keep living in a delusional world of download numbers = market and JMX wil save the world. (Yes WAS 5 admin objectsare bult with Mbeans)

    I am not bashing the products, just defending ours
  25. very clear?[ Go to top ]

    WLS's announced 8.1 release did not even tout J2EE 1.4


    The J2EE 1.4 spec is not baked. Sun imposes
    strict requirements on J2EE licensees with respect to
    to J2EE compatibility. For instance, I don't believe
    a J2EE licensee can put J2EE 1.4 features into a J2EE 1.3
    product.
  26. very clear?[ Go to top ]

    <matt>
    Funny that niether IBM nor JBOSS have a shipping SOA product even closely comparable to BEA's WebLogic Platform
    </matt>

    Please keep the discussion real. BEA WP is a buggy piece of crap. The latest 8 release is a desperate move to try and appear different when most of it is expensive garbage.

    <matt>
    J2EE is not going to die, but products that are stuck at ONLY implementing J2EE might -regardless of whether they are commercial OR open source
    </matt>

    I think what ben (who appears to be JBoss's sales director) is saying is that JBoss was ahead of the curve in offering services based aspect oriented POJO work. JBoss is innovative. If JBoss 4 is really EJB applied to POJO it will be worth a lot.
  27. very clear?[ Go to top ]

    It would be nice to get into a discussion of the impact of services development without the IBM, BEA, and JBoss salespeople getting into a flame war about their products. As a developer, I'm interested in the implications of a vendor's technology directions, but not from the standpoint of a prospective buyer. I'm curious about how the industry is going to solve old problems, how new ways of doing things will be discovered, and how the industry as a whole will evolve. If I wanted to listen to vendors fight, I'd go elsewhere.

    Quite frankly, none of the J2EE vendors offers a perfect solution. They all have problems, from WS to WLS to JBoss. Each one has strengths, too. If I am to listen to the strengths and weaknesses of each, I'd rather it be in the spirit of spreading knowledge, developer to developer. Warnings are appreciated and helpful; flames are neither.

    So please, if you're a sales director trying to convince people about how great your server is, post elsewhere. What could have been an interesting discussion about services-oriented architectures has devolved into a pissing match between people who don't have to use this software every day.

    caveat: at least the JBoss guy had some interesting things to say in the context of his sales pitch, and at least he was honest about making a pitch.
  28. Sorry....[ Go to top ]

    Sorry about hijacking a discussion. I thought I was starting some good high level talk about the value of servies based architecture and not a pissing contest. I tried not to put down any product in my initial post and respond to some of the posts about J2EE going away.

    I do strongly belive in a services based future and I think the answer is very simliar to the design of operating systems. Afterall, a container should be the next OS, the enterprise OS. All good OS are based on a set of services that run in a kernel.

    If the Kernel is modular ala Linux more the better. This can be achieved in a variety of ways in Java, some use JMX.
  29. Sorry....[ Go to top ]

    If the Kernel is modular ala Linux


    tanenbaum oh, tanenbaum...
  30. Good point[ Go to top ]

    tanenbaum oh, tanenbaum...


    Whaaaahahahahaha. Good one :-)
  31. Reference[ Go to top ]

    In case you're wondering about the Tanenbaum reference, here is the original post (I believe).
  32. SOA and its meaning[ Go to top ]

    Aren't we talking about different things here? I really like JBoss's JMX architecture, but if you call this an SOA, I think you are talking about something different than a high-level architecture that views application functionality in an enterprise as a set of services. Similar idea, but not just a different scale.

    Or am I missing something?
  33. SOA and its meaning[ Go to top ]

    No, you are not missing anything. SOA generally refers to higher level Provider/Broker/Requestor architecture, not lower level implementation details (which might designed in a services oriented way). In fact it is all about masking the implementation architecture from the requestor of a service.
  34. POJO on Steroids[ Go to top ]

    <Brian>
    If JBoss 4 is really EJB applied to POJO it will be worth a lot.
    </Brian>

    ...I guess you'll at least get what you pay for...
     ;-)


    Actually, this is one of the points of moving beyond ejb- To simplify enterprise app development so that it is as SIMPLE as possible -even simpler than POJO.

    The full complement of integrated frameworks such as this -based on standard components and a single server - are what make up the platform.

    I have personally seen shops experience VERY REAL 6-fold productivity increases with the platform!
    Matt
  35. Tuxedo, Encina, and Top End, as well as IBMs' CICS environment have provided services-oriented architectures since the 80s, albeit on a largely proprietary level. (however, BEAs' Tuxedo connects seamlessly to IBM CICS and both support remote 2PC). The advantage to the new SOA is simply Java, J2EE, the web environment itself, and the standards that accompany Web Services. Before J2EE, services were almost mandated as stateless with the real program logic developed in one of the OS-bound languages. With J2EE, we have the first opportunity of the virtual distributed environment and the programming logic itself can live entirely in the virtual tier, in a single language, sharing resources across all machines. BEA ported the set of distributed Tuxedo capabilities in implementing the J2EE tier, thereby extending the distributed science that began some time ago. IBM points out with every presentation that WebSphere is a brand, not a product, and JBoss is doing an excellent job of providing a platform that evangelizes and explores the limits of J2EE. Regardless of the flavor, we are all proving that an open J2EE-based infrastructure is a viable, production-ready, performant architectural option and we are winning the architectural battle in the absence of a released and viable platform from Microsoft.

    Every large transaction facility in the world (representing the most demanding Reservation, Banking, and Financial applications world wide), has in the middle tier an SOA, they are just not all (yet) based on Java.
    David Ressler, BEA
  36. Service Based Architecture?????[ Go to top ]

    Isn't it was solved by Corba or Jini years ago?
  37. Indeed, SOA is not tied to a particular programming language or operating system/platform.

    The difference now is the move from high-end proprietary stand-alone products to enterprise-scale SOA based on industry standards (e.g. BPEL) that's available to the mainstream (i.e. developer skills availability and ubiquitous deployment infrastructure).

    Doron\

    Reliable standards-based orchestration
    Download Collaxa 2.0 (www.collaxa.com)