IBM unveils new WebSphere product line

Discussions

News: IBM unveils new WebSphere product line

  1. IBM unveils new WebSphere product line (37 messages)

    IBM unveiled its latest set of service-oriented architecture offerings yesterday, filling out its WebSphere product line with a lightweight, Java-based enterprise service bus (ESB) and a Business Process Execution Language (BPEL) engine. IBM also announced upgraded offerings for WebSphere Message Broker and its business modeling and monitoring tools.
    New and upgraded offerings in the IBM SOA Foundation include:

    -- The new, Java-based WebSphere Enterprise Service Bus.
    -- The new WebSphere Integration Developer, an Eclipse-based tool for the development and integration of composite applications.
    -- The new WebSphere Process Server, a BPEL engine that coordinates business processes with the new ESB, is built in to provide the messaging backbone.
    -- The new WebSphere Everyplace Deployment, which monitors events in the application layer and sends alerts out to XML edge devices.
    -- The upgraded WebSphere Message Broker.
    -- The upgraded WebSphere Business Modeler, which allows business processes to be modeled prior to development.
    -- The upgraded WebSphere Business Monitor, which provides a dashboard view of Web services performance.
    Read IBM trots out muscle-bound SOA

    Threaded Messages (37)

  2. SOA overdose[ Go to top ]

    Regarding Dave Chappell's comment in the article about the WebSphere "Sargasso Sea of technology": he has a point. Making sense of and evaluating all of these products (in addition to all the pre-existing WebSphere products of course) is not for the faint of heart. Didn't someone say that SOA would simplify IT and better align it with the business ? The jury's still out on the "simplifying" aspect...

    I can't resist throwing in a ridiculous conspiracy theory: the whole of WebSphere was invented by IBM in order to convince customers that they should stick to mainframes, OS/390, Cobol, CICS and DB2 :-)
  3. SOA overdose[ Go to top ]

    Didn't someone say that SOA would simplify IT and better align it with the business ? The jury's still out on the "simplifying" aspect...

    Now we're all holding our breath to see IBM's open source ESB!
  4. IBM unveils new WebSphere product line[ Go to top ]

    I am curious, who actually likes to work with ibm stuff, I mean it hardly works. I don't understand how they can build such a great product as eclipse and at the same time deliver websphere/rad wich is IMO a big bunch of pathetic crap. I dont even care to look at the other pathetic crap.
  5. I am curious, who actually likes to work with ibm stuff, I mean it hardly works. I don't understand how they can build such a great product as eclipse and at the same time deliver websphere/rad wich is IMO a big bunch of pathetic crap. I dont even care to look at the other pathetic crap.
    +1.

    Websphere 6.0 is a joke and so is RSA.

    IBM seem to go out of their way to make JEE more complex than it needs to be. Ever tried to use their appliaction server without their IDE.

    Talk about Flaming Hoops.
  6. IDE 6 Flaming pile of[ Go to top ]

    Flaming Hops have you tried using their IDE for 6, talk about a flaming pile of ......
  7. Do you mean RAD?[ Go to top ]

    Have you used RAD? Have you used WebSphere for implementing SOA? I've been using WebLogic and it's pretty nice, although it's not without its problems. Seems like performance for us is the most overriding issue.

    I was thinking of looking into RAD for SOA because in my previous experience, you couldn't really work with WAS effectively unless you used WSAD. Any thoughts? TIBCO is an option too!

    Lou
  8. IBM unveils new WebSphere product line[ Go to top ]

    Websphere 6.0 is a heckuva lot better than 5.1.1.1.1.1. I couldn't even get the latter to work on Linux whereas 6 installed quite smoothly.

    It's not an app server for small, lightweight development, that's for sure. If you are writing straightforward webapps, Websphere is severe overkill.
    Websphere 6.0 is a joke and so is RSA. IBM seem to go out of their way to make JEE more complex than it needs to be. Ever tried to use their appliaction server without their IDE. Talk about Flaming Hoops.
  9. IBM unveils new WebSphere product line[ Go to top ]

    I am curious, who actually likes to work with ibm stuff, I mean it hardly works. I don't understand how they can build such a great product as eclipse and at the same time deliver websphere/rad wich is IMO a big bunch of pathetic crap. I dont even care to look at the other pathetic crap.
    +1. Websphere 6.0 is a joke and so is RSA.IBM seem to go out of their way to make JEE more complex than it needs to be. Ever tried to use their appliaction server without their IDE.Talk about Flaming Hoops.

    Wonder if you guys are speaking from "really real" experience's. There are lots of real life case studies, where WebSphere is being used without WebSphere/Rational IDE.

    J2EE can be as complex as want it to be and at the sametime it can be as simple as you want it to be. WebSphere just like others, is just a J2EE server.

    BTW, do you mind sharing your experience with WebSphere/RSA, I can try and help you out.

    Kapil Isarni, IBM
  10. WebSphere just like others, is just a J2EE server.BTW, do you mind sharing your experience with WebSphere/RSA, I can try and help you out.Kapil Isarni, IBM

    I do speak out of "really real" experience, as I am a couple of months dedicated deployment manager in a large project that makes use of WebSphere.
    Yes as a J2EE server WebSphere is indeed just like others, but not where it goes about deployments.
    It seems like the whole point of it is to make it tightly coupled with retarded IDE and to make the whole thing so complicated that you would definitely need to hire couple of IBM consultants to solve that mess.

    To pretend I'm constructive, I'll write my personal complaints top 5 list:

    1. Deployment process is made unnecessary heavy. To stop-undeploy-deploy-start an application, it takes about a minute, compared to a few seconds on almost any other J2EE server.

    2. WebSphere config repository breaks on numerous occasions (e.g. when someone using web console while JACL script is running). The application is there in list but it can't be uninstalled, because it's not installed. No it can't be installed either, because it's there in list. After installation, the application can't be started. Because it's not installed. Wait a minute... no, it's installed already. And indeed no documentation is available on the internal repository XML format. Therefore, the problems with configuration repository are often only cured with total reinstallation of WebSphere.

    3. IBM assumes that a developer is always working alone and with WSAD. With these ugly wizards. Hey IBM, have ever heard of dedicated build servers, automated builds/tests, continous integration, and so on? Obviously not. Building and deploying an EAR from WSAD without WSAD itself is obviosly not what IBM want to support. The possible solutions for those creative ones are:
    a) rewrite the build process with ANT or
    b) use that freaking 'headless WSAD' build... oh yes you'd then need one more WSAD license for build server as well

    4. XMI format for IBM-specific extensions is not documented. I have to use that thing, WSAD, you know, just to make things like transaction attributes of datasources look properly in XMI.

    5. Logging of error messages. That's a plague that comes over absolutely all IBM products, from OS/390 and OS/400 till WebSphere. If a JACL script fails, it says something very informative like 'a quote is missing'. No line number, though. Go figure. If there are incorrect user rights on one of WAS directories, the error message is so vague that surely only IBM consultant can find the cause. I don't want even start on classloaders and so on... obviously you at IBM rarely test with more than one application on the server.

    * * *

    At this moment, I'd _never_ would use WebSphere as an application server, unless the customer wants me to.
    But the big accounts seem to like IBM and they buy everything IBM sells. Today IBM sells J2EE and Struts, they buy J2EE and Struts. Tomorrow IBM will sell Spring and AJAX, they'll buy it as happily as they can - for no one can be fired for buying IBM.
  11. ...and by the way, the whole project (~>100 man-year) was originally started to port the applications from WebSphere 3.5 to WebSphere 5.
    Because, guess what? Because these applications for WebSphere 3.5 were made with certain IBM extensions which were not supported anymore in WebSphere 5!

    And I guess that with advent of new WebSphere product line, even more customers will hopefully become hooked to happily use on average ~5% of that behemoth's capabilities, until one day it becomes obsolete and new upgrade is inevitably required.

    What a brilliant marketing strategy!
  12. no bashing here[ Go to top ]

    I'll write my personal complaints top 5 list:1. Deployment process is made unnecessary heavy. To stop-undeploy-deploy-start an application, it takes about a minute, compared to a few seconds on almost any other J2EE server.

    hmmm, i am sorry to not join the IBM bashing folks here. To me BEA is not any faster when it comes to deployment. Perhaps its even slower. The whole deployment issue is really one of the strong arguments for leightweight J2EE approaches.

    Marc Logemann
    http://www.logemann.org
  13. no bashing here[ Go to top ]

    I'll write my personal complaints top 5 list:1. Deployment process is made unnecessary heavy. To stop-undeploy-deploy-start an application, it takes about a minute, compared to a few seconds on almost any other J2EE server.
    hmmm, i am sorry to not join the IBM bashing folks here. To me BEA is not any faster when it comes to deployment. Perhaps its even slower. The whole deployment issue is really one of the strong arguments for leightweight J2EE approaches.Marc Logemannhttp://www.logemann.org

    A one-minute deployment is lightning fast. I have to use a really bad B2B/SOA product where deployments take from 30 minutes to 2 hours on dev, test, and prod. It also has a space leak so deployments get slower and slower as the OS starts thrashing once the server needs more than 1 or 2 GB of memory. You have to undeploy, and delete all your apps to clean up and lose all the configuration which takes hours to setup and save.

    Sorry to get off topic but I don't feel sorry for people complaining about one minute start-up times.

    In any event, back in the good old days I was involved in an evaluation of J2EE app servers and BEA was a joy to use over our SilverStream/Novell piece of crap. Gotta love hot-deployments for development. Even JBoss wasn't too painful, despite the lack uof usable documentation. Actually, you could use the BEA documentation to help with JBoss. The guy doing the evalaution for WebSphere was pulling his hair out and cursing all day long.

    Actually, they tried to make me use WSAD on my current job, even though we don't do J2EE on this team, and I've decided they should just drop the 'W' and call it SAD. ANGRY would be a better name, but SAD's good enough. I finally just downloaded Eclipse. It's funny how much better the OS project made it. Still has that dumb-ass problem where a non-perfect close destroys your workspace though. I hear the newer version of SAD is better but I'm 'once bitten, twice shy', if you'll forgive the cliche.

    Anyway, I digress.
  14. 1. Deployment process is made unnecessary heavy. To stop-undeploy-deploy-start an application, it takes about a minute, compared to a few seconds on almost any other J2EE server.

    Agreed. It's only that it takes 15 minutes for our dedicated box to perform the task, not 1 minute ;)


    >>>2. WebSphere config repository breaks on numerous occasions (e.g. when someone using web console while JACL script is running). The application is there in list but it can't be uninstalled, because it's not installed. No it can't be installed either, because it's there in list.

    "Childish" WAS problems. A rule of thumb in my department is to _never_ ever use the WAS admin console of the build server. It's a one-time administrative task performed during setup and nothing else.

    >>>3. IBM assumes that a developer is always working alone and with WSAD.

    Absolutely :)) And they are naive to really believe in benefits it provides. Our customer forced use to use WSAD at first (and drop the favourite IDEA). I will just quote on of the post I saw here on the site: "PRODUCTIVITY DISASTER". Luckily, we have built enough trust over 1 year to let us use the tools _we_ consider productive. Now, IDEA + Ant + CruiseControl, and nobody can match that speed on customer's side. Much closer to a "rapid" development, if its applicable to WAS at all.

    >>>4. XMI format for IBM-specific extensions is not documented. I have to use that thing, WSAD, you know, just to make things like transaction attributes of datasources look properly in XMI.

    That one is really astonishingly stupid on their side. Even their official InfoCenter docs tell you to use WSAD for these kind of things. TIP: they use the XML ID fields, so if you rename those senseless numeric timestamp values to _your_ values (say, meaningful strings and names), you can match them in EAR descriptors and IBM XMI binding files. I did that to automatically map bind application users to the ones specified in a custom auth provider for WAS.

    >>>5. Logging of error messages. That's a plague that comes over absolutely all IBM products

    Oracle takes the same route, so we have a trend here :)
  15. I can take a stab at some of your complaints with our experience RE: WAS 5, we are about to migrate to WAS 6

    1. You don't need to stop, uninstall install restart to deploy your app you can use the update command, to refresh an ear, although expecting it to update in seconds is optimistic. Of course this will only be valid as long as the binding info (groups to roles, db etc) has not changed, in which case you will have to uninstall and reinstall your app. Alas I think the JBOSS publishing to a folder hot deploy mechanism is and will remain just a dream. We do this update thru a jacl script

    2. Yes, we've seen this and worse. In one case the jacl script manage to corrupt the entire repository. This was tracked down to a bug and the apporiate fixpack was applied.

    3. IBM push headless ant with WSAD, IMHO it is not necessary . I agree that developers should not be triggering builds and deploys from their local machines or sacrificing a licence, a wonderful change control process which would give our admins a fit. WAS (Server) includes a special version of ant (WS_ANT) which allows you to link of to some specialized ant tasks to do the EJB deploy and publish, if anyone has gotten JSP compile to work i'd love to hear from you. So you can constuct a build script and run it on a build server (although this must have the WAS libs and tools).

    4. Yup, I agree and you wonder why 3rd party tools like xdoclet are always a version or 2 behind for deployment descriptors..... and also why you need to develop with RAD or WSAD.....

    5. Yup, jacl is the pits..... and error messages are always especially helpful (NOT). Although you now have the option of JYTHON as of 5.1, not that it probably matters. Classloaders don't get me started...... a constant never ending battle.
  16. IBM unveils new WebSphere product line[ Go to top ]

    Alas I think the JBOSS publishing to a folder hot deploy mechanism is and will remain just a dream.

    I think this what frustrates a lot of people who use WebSphere. In websphere 6.0, they now have a separate tool that you have to run in order to get hot deployment. But it is still bad and require wierd ass configuration. Why a seperate tool? In Jboss, you just dump a ear file or a sar file and you're off and running. No such luck with Websphere. It is just like their other products such as lotus notes and clearcase: big, memory hog and a lot of annoying things that makes the user's life difficult.

    It took until version WAS 5.1 until our product was stable. We made the mistake of using too many J2EE components. And yes, at our company we have websphere because of management. I've been in this field for some years now and I have never heard a positive comment about Websphere.

    Nice fact: to install Websphere 6.0 you need 2.2 Gig. The developer studio: 1.8 Gig.

    Dino
  17. Development or Production[ Go to top ]

    I am not a big WebSphere fan (using it since version 3.0) but it is not as bad as people make it in this thread.

    My experience is that WebSphere is not a good choise for development (especially if your are not using WSAD or RAD) because startup time, (re-)deployment takes to long and the configuration part is too complicated.

    In many projects we are using Tomcat or JBoss (depending if we needed EJB or not) in development and WebSphere in production. WebSphere in production is a good choice because of a professional support (that is even cheaper than JBoss support) and a good manageability of big clusters and lots of servers. (Deployment Manager) And if you look at monitoring tools a lot of vendors support just WebSphere and Weblogic...

    Mirko
  18. Development or Production[ Go to top ]

    I am not a big WebSphere fan (using it since version 3.0) but it is not as bad as people make it in this thread. My experience is that WebSphere is not a good choise for development (especially if your are not using WSAD or RAD) because startup time, (re-)deployment takes to long and the configuration part is too complicated. In many projects we are using Tomcat or JBoss (depending if we needed EJB or not) in development and WebSphere in production.

    you don`t think it's risky to do development on a completely different platform than you run on in production? I assume you go through a prod-like test environment before deploying to production.
  19. Wonder if you guys are speaking from "really real" experience's.

    No I work at home with it because i like it so much, common man is that all you can say about it. How real is real to you? If I am working on a team with 4 other people i have no right to say anything because i am not real. I worked for banks only the last couple of years, is that real enough? And I can tell you that only the board still believes in IBM, if you lose them your gone man. I take any chance I get to convince 'm to dump it forever.

    some reason's why.
    - No console output in debug mode
    - stacktrace elements not clickable anymore
    - server starts in debug mode if you want to just start it.
    - servers just starts in non debug mode if you want to debug (there is unfortuannaly no pattern in this, just random)
    - Dont ever remove or rename anything, you will spend hours on cleaning garbage.
    - struts editor cannot save files. IDE says it is saved, close file , open file again , changes lost.
    - stacktraces from yesterday pop up making you think something is wrong now.
    - stopping the server takes minutes???, starting it up also.
    - custom tag assist sometimes works.
    - html assist sometimes works.
    - html assist (if works) totally ***** up indentation of the whole page.
    - the server hangs for minutes several times a day.
    - j9 cannot run in debug mode with crashing.
    - don't get me started about clearcase that comes with it.

    I could go on for a while. If i would deliver like that i am out of a job real fast. Ibm used to have my sympathie because of eclipse and other open source initiatives but that is fading quickly now. They seem to be intersted in marketing only these days. Just sell the blue boxes to big companies letting us work with it. Well i'am not going to work with it anymore after i finish my current job, I' am going to explicitly look for non-ibm jobs. It-Market is to good these days to just take anything that comes along. And then you will never hear from me again.
  20. RADv6.0 and its initial fixpacks 6.0.0.1 etc, all use a quite old version of the IBM JVM to run under. This JVM had issues such as throwing GPF faults from the J9 JIT during debugging.
    The RADv6.0.1 update upgrades the JVM to a more recent version which fixes some of the instability issues.

    One of the biggest problems with RAD is its resource usage. I run with 1GB of RAM and that is pretty tight. I cannot ran too many other applications at the same time. 2GB is recommended. Part of this is due to memory leaks in areas such as the JSP editor. I have seen the memory used leap from 196MB to 600MB after editing a couple of JSPs. Using Open With...=>Text Editor helps this issue.

    As for WASv6, I have found it to be pretty solid. Certainly a lot better than the previous releases.
  21. I have been following this post after I added my brutal comment. There are 17 explicitly negative comments and 4 rather poistive. Are Matt and Kapil still there, do you learn anything from this or are you just offended? What have you got to say about this apart from calling me a basher or not "really real". If I add a nice diplomatic comment about how hard it is to work with IBM software then that would not reach you at all. I am brutal because it is the only way to get through, do something about it please. The stuff works for 98% but that last 2% gives a lot of people a really hard time. Either you don't take/get the time or you are not up for it. I believe it is the first rather then the latter.
  22. I, for one, am no lover of IBM (I've had my own share of frustration), but I have to say that I'm impressed with RAD/RSA and WebSphere for the kind of work I do in large organizations (banks, insurance) which require a lot of integration.

    By the way, I disagree with the comment about WAS 3.5 to 5.1 migration. I've recently completed a project to migrate all of an insurance companies apps from 3.5 to 5.1 and they migrated without any code changes at all !! Only had to repackage the WAR file to be compliant.

    Gordon
  23. 1. My company is one of the first customer who bought the entire suite of IBM New product line based on SOA. So far not even a single day we have gone home with peace of mind. 2. You talk about any thing from RSA,RAD6.0,WID60,6.0.1, ProcessServer 6.0,6.0.1. None of these would install properly on windows server or professional editions after applying all the service packs on the OS and also the latest fix patches on the products. 3. None of these will successfully install test environment properly. No proper customer support. WE have lost all the UML Models developed on RSA when we applied the fix patch of 6.0.1. 4. No body at IBM is bothered about the known issues list and we have even raised a ticket, but its of no use. 5. HOw can a product fix pack does not have a backward compatibility. we have sent the .emx files to IBM, but still they could not find. 6. RSA is supposed to be supporting UML 2 version, where as we are not finding many notations in the activity diagrams and Sequence Diagrams. 7. Latest suite of websphere does not support EJB 3.0 which is already an year old. 8. It also does not support struts 2.0. but we can do a work around for this. I think i have mentioned enough issues to say the product is not well tested. IBM thinks that and can sell any product to their clients like us.
  24. I went to a WebSphere Portal installation/configuration/training... presentation few months ago. This beast is a nightmare to install and configure. The guy who made the presentation said that yes, it's complicated, , but it permits to company like the company I'm working for (big consulting company) to bill more hours to our clients. Pathetic!! So now, I understand the philosophy of IBM company and I'll never recommend anymore their products to our clients!

    (sorry for my poor english)
  25. I really hate IBMWebsphere. Why? for a simple deployment of an ear file into the server , without using RAD, I was supposed to write a build file which does that. In weblogic the ant tasks provided by the server is smooth like silk and the deployment happened in secs. BUT for the past 4-5 days, I am still trying to figure out which monkey will make this work on websphere 6. I came to know that there is a task but the documentation is so less on net, and ofcourse the IBM Red yellow green books....i dont know whether they know how to writ documents...I am still figuring out. One thing is for sure....IBM makes products COMPLEX so that THEIR PERSONNEL (pseudo consultants) can only decrypt the CIPHER in it planted by themselves..... I am really feeling like kicking this app server out of business. Just imagine I had to include ejb stubs separately as it was getting called from a POJO and the ejb was in a different ear. I am using Jbuilder 2006 , and JBUILDER is perhaps the best IDE in market today better than IntelliJ . Even their product eclipse....is what i believe to overshadow the good job done by SUN Microsystems. SUN - eclipsed ... get the idea. Never make their own product..They brought Rational....IBM keeps on buying companies...HA HA... Actually IBM is... International Bullying Machines.... I hope that really help the suckers...
  26. radwhich hmmm, hmmm[ Go to top ]

    I am curious, who actually likes to work with ibm stuff, I mean it hardly works. I don't understand how they can build such a great product as eclipse and at the same time deliver websphere/rad wich is IMO a big bunch of pathetic crap. I dont even care to look at the other pathetic crap.

    Something about a websphere radwhich stuck me a funny.
    OK not that funny.
  27. radwhich hmmm, hmmm[ Go to top ]

    I am curious, who actually likes to work with ibm stuff, I mean it hardly works. I don't understand how they can build such a great product as eclipse and at the same time deliver websphere/rad wich is IMO a big bunch of pathetic crap. I dont even care to look at the other pathetic crap.
    Something about a websphere radwhich stuck me a funny.OK not that funny.

    What is radwich? Maybe i can laugh to.
  28. WebSphere Everyplace Deployment[ Go to top ]

    Your short tag line for WebSphere Everyplace Deployment is a little miss leading, WED is a product built ontop of Eclipse RCP that extends the SOA programming model and J2EE APIs to the desktop. It allows Smart Client applications to be developed and for them to be integrate into an SOA infrastructure using WebServices and Messaging etc.It can also be used to offer an integration point at the glass using Eclipseo mix, SWT,Swing,Browser and ActiveX application all within one user experience. The WED client will be released on Windows and Linux operating systems. It also includes a server component that runs ontop of WebSphere that allows application to be deployed out to large number of client desktops.

    Not sure where you got the monitoring bit, may just have been a cut and paste issue.

    Good to see the IBM bashers are out early on this post !!

    Matt Perrins
    IBM Software Group
  29. WebSphere Everyplace Deployment[ Go to top ]

    Good to see the IBM bashers are out early on this post !!Matt PerrinsIBM Software Group

    Is not a bit easy to just call me a basher. Have you got any idea how frustating it is to have to work with stuff that does not work. Why is there not going any effort into supplying some meaningfull messages when something is wrong.
  30. WebSphere conspiracy theories[ Go to top ]

    Good to see the IBM bashers are out early on this post !!
    Actually it was late in my case as I'm in old Europe :-) But anyway, I don't consider myself an IBM basher at all and I posted my conspiracy theory solely as a joke. As was predictable, other theories popped up afterwards and their authors seem dead serious about them. But these theories sound rather naive about the way large organizations work. Do people really believe that there are big secret meetings among IBM's top software honchos with agenda: "How to make our products even more complex so that we make a fortune on consulting" ? Mmmmm... I think a more reasonable explanation is that IBM, as a huge company selling everything from hardware to software and services to countless other huge companies has to cover requirements that go all over the map. They have an applications portfolio made of thousands of products with various origins, many of which came through acquisitions. And somehow they must unify all of that into a cohesive, all-embracing platform and vision. Given the complexity of the problem, it's somewhat miraculous that they manage so well. But the downside is that, from the perspective of a single customer, making sense of that behemoth platform, using some parts of it and keeping somewhat up to date with them takes vast resources. Large system integrators (among which IBM G.S.) thrive on this complexity and are usually keen on dragging the customer ever further on the complexity path because this will tighten their control on the customer and lead to future budgets. SOA is widely publicized as a way to "simplify" IT but as I said yesterday, the jury's still out on that one! People's interest in so-called "lightweight" solutions such as Spring/Hibernate/Tomcat/Axis/etc is quite often, in my humble experience, a consequence of being burned by the complexity of more mainstream, vendor-approved solutions. But are these lightweight thingies going to make that much difference in the long run ? Again, the jury's still out, but I'd say it will only be incremental improvements. For more profound changes, we shall need something radically different from all this J2EE "legacy" stuff :-)
  31. my view[ Go to top ]

    Alain is correct. Its ridicolous to think that the websphere unit is creating websphere with complexity and non-usability in mind to sell (a) more boxes and (b) more consulting. At the end, they are competing with BEA, JBoss and even with Tomcat. And dont forget that IBM is not that small, every unit has its own revenue targets and stuff. I dont think that the websphere product manager can report to Sam that his product is absolutely useless, but the service unit sold 3% more consulting days last year. Its not the way any company that size works.

    And products like WSAD are not that bad. I know many developers out there who were not able to develop webservices without it. This doesnt necessarily mean that the developer is incompetent but perhaps he has a different focus. It really depends on the target audience. We, as we write here on TSS, are not the average developers. The typical developer working in big companies hardly knows TSS, nor does he consume the latest and greatest open source frameworks. Talk about Spring at companies like Vodafone or DaimlerCrysler, you will need some time to meet someone who has experience with it (if you find one at all).

    But one thing is for sure. Application Servers nowadays are monsters. And when you look at all these specs they have to implement (plus the ones which they add as value add like SOA things), its no surprise that deployment takes plenty of minutes (the one minute complaint was a joke wasnt it? We have about 5 minutes minimum deployment time here). As someone allready said, one should really carefully judge if its necessary to use all the bells and whistles. Currently i am working for one of the biggest telcos in the world and even here, there are scenarios where we could live quite happily without the full J2EE stack, and i speak about a CRM system with some traffic.

    Marc Logemann
    http://www.logemann.org
  32. my view[ Go to top ]

    Its ridicolous to think that the websphere unit is creating websphere with complexity and non-usability in mind to sell (a) more boxes and (b) more consulting. At the end, they are competing with BEA, JBoss and even with Tomcat. And dont forget that IBM is not that small, every unit has its own revenue targets and stuff. I dont think that the websphere product manager can report to Sam that his product is absolutely useless, but the service unit sold 3% more consulting days last year. Its not the way any company that size works.

    I agree. No product manager would ever agree to a strategy of deliberately making a product difficult to use. The complexity is there because it's both a broad and deep product.

    I've worked in a large company (100K+ employees). Different business units work with each other ony up to the point that it makes sense. The WebSphere product group would not hesitate for a second, for example, to use external resources over internal resources if it that made more more.


    PJ Murray

    CodeFutures Software

    Code Generation for Java Persistence
  33. WebSphere conspiracy theories[ Go to top ]

    Good to see the IBM bashers are out early on this post !!
    Actually it was late in my case as I'm in old Europe :-) But anyway, I don't consider myself an IBM basher at all and I posted my conspiracy theory solely as a joke. As was predictable, other theories popped up afterwards and their authors seem dead serious about them. But these theories sound rather naive about the way large organizations work. Do people really believe that there are big secret meetings among IBM's top software honchos with agenda: "How to make our products even more complex so that we make a fortune on consulting" ?

    I doubt it's a conscious decision. I don`t know the real reason. Incompetence? Management interfering with the design to ("make WSAD a neccesity")? There are clearly serious deficiencies in the product. Have they implemented any sort of XDoclet or annotation support yet?

    Given my experience with some consultants lately, I can see them recommending the most complicated and time-draining app server.
  34. WebSphere conspiracy theories[ Go to top ]

    <blockquoteI think a more reasonable explanation is that IBM, as a huge company selling everything from hardware to software and services to countless other huge companies has to cover requirements that go all over the map. They have an applications portfolio made of thousands of products with various origins, many of which came through acquisitions. And somehow they must unify all of that into a cohesive, all-embracing platform and vision. Given the complexity of the problem, it's somewhat miraculous that they manage so well.
    I still fail to see what problems like obscure error messages, failing config repository, WSAD as the Only Way of development, and lack of documentation have to do with the reasoning stated above.

    What I do see is that the reason IBM doesn't care is that it sells not to actual developers (as, say, IntelliJ does), but to top-executives, who never will find themselves in front of WSAD eating their workspace or struggling with cryptic error messages. Top-executives, on the other hand, do like sentences like "a BPEL engine that coordinates business processes with the new ESB, is built in to provide the messaging backbone", no matter what it means really. And remember, no one's fired for buying IBM... even if it was WebSphere 3.5.
  35. The whole point of ibm software ...[ Go to top ]

    Speaking from experience, the whole point of ibm software is to sell big expensive boxes and to get dumbassed ibm consultants out at a thousand bucks a day to make it work.
  36. Re: IBM unveils new WebSphere product line[ Go to top ]

    What will IBM and BEA do when businesses start realizing the benefits of lighweight containers such as Spring? 99% of the applications being developed today DO NOT need the complexities of EJB, etc. that tie them to an application server such as WebSphere or WebLogic. For those wondering, yes, I have developed my share of EJB's in the past and as a consultant have developed applications for major companies (American, Verizon, FedEx, Southwest), so I do speak from experience. My advice, take a hard look at what you are really wanting/needing those "heavy" app. servers for and decide if there is a better way to architect your system. Do you really need remote deployment of EJB's? Do you really need some RAD tool (provided by the big company) to handle your web services deployment files?

    I will end by saying that I know the major (expensive) app. servers provide more than just EJB capabilities, but are you really using these services? Just recently the IT industry has begun a monumental move towards web servics and SOA, however these app. servers are not necessary for this type of development either.

    Remember, big companies with expensive app. servers MUST make money for their shareholders, so what do you think they advertise regarding their products: advanced, complex architectures that support the need to buy their products.

    Just my $.02
  37. Spring[ Go to top ]

    Shaun, note that BEA seems to have done so - they and Interface21 announced a formal association of sorts (covered on TSS, even - aren't you reading?) back at JavaOne.
  38. They will follow the tides[ Go to top ]

    Mark Carges, the CTO of BEA systems, gave a speech (http://dev2dev.bea.com/pub/e/705) at this year's Java One talking about Spring and other lightweight containers. He pointed out that there were indeed complexities with EJB development that Spring and other lightweight containers could remedy.

    Mr. Carges pointed out that Spring and other lightweight open source projects can be readily deployed on BEA's latest app server, WebLogic Server 9.0, and that the next revision of BEA's Eclipse-based Workshop developer tool will provide tooling support for those projects/frameworks. So why would you want to host Spring on WebLogic or some other enterprise-grade app server? The same reason you would host an EJB application -- high performance, clustering, scalability, security, availability, support, etc.

    There are plenty of reasons that a company would choose to use open source technologies on top of vendor's app servers. Most app server companies do not consciously try to tie a developer to their platform. It is simply a reality that the J2EE/EJB specifications do not define every last function of an app server. It is therefore up to the app server vendors to fill in the holes with custom, or proprietary, functionality. I totally agree that some app server vendors go out of their way to make things overly complex, but not all of them do.

    I would further point out that BEA has taken a large chunk of it's proprietary functionality and moved it into open source projects (Beehive and Pollinate). IBM has made similar donations over the years. Believe it or not, vendors like to compete on how well they solve business problems, not how tightly they couple their customers to their technology. OK, I will make a slight concession on that point for our favorite Redmond company. But they don't have a Java app server, so my point still stands!