Gartner: IBM takes #1 Appserver marketshare position from BEA

Discussions

News: Gartner: IBM takes #1 Appserver marketshare position from BEA

  1. IBM has taken the No. 1 appserver market position away from rival BEA, according to a study to be released this week by Gartner dataquest. The study pegged IBM at 37%, BEA at 29%, with Sun and Sybase a distant third and fourth. "IBM gained market share of new-license revenue at the expense of BEA and others."

    Read IBM pulls away in app server race.

    Threaded Messages (102)

  2. BEA is at 29%

    <quote>
    "We thought (BEA's) WebLogic was more of a product than a marketing construct" versus IBM, said Mike Parks, chief information officer of Virgin Mobile USA.
    </quote>

    BEA at the crossroads
  3. Until IBM actually breaks out the units and revenue for WebSphere itself, there is no possible way that they are ahead. The fact that they hide their numbers (as explained in other published reports on J2EE server marketshare) is the best possible indicator that they remain well behind WebLogic. It does seem that IBM has established a definite #2 position though, by convincing a good number of ISVs that they have to support WebSphere in addition to the default support that ISVs have traditionally provided for WebLogic. It's an excellent strategy, but it's not been enough to get them anywhere close to WebLogic in terms of ISV support, actual deployments, and developer mindshare.

    Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone. ;-)

    Peace,

    Cameron Purdy
    Tangosol, Inc.
    Coherence: Easily share live data across a cluster!
  4. 7.1 billion downloads[ Go to top ]

    Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone.

    Good one, Cameron :)
  5. 7.1 billion downloads[ Go to top ]

    Does anyone heard some business solution based on JBoss?
  6. 7.1 billion downloads[ Go to top ]

    We did try a solutions in JBOSS. But there are so many optimistic and pessimistic locks in there ..and beleive me for no reason ...pain in the neck..we had to revert back to weblogic where we had make a roolback on so many things....
  7. Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone


    How is this possible ?
    World population as of 07/01/03 is 6,302,309,691
    http://www.census.gov/cgi-bin/ipc/popclockw
  8. Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone

    >
    > How is this possible ?
    > World population as of 07/01/03 is 6,302,309,691
    > http://www.census.gov/cgi-bin/ipc/popclockw

    Obviously some of them downloaded JBoss copies for the next generation before it becomes fee based ;-D
  9. 7 billion, eh?[ Go to top ]

    lol...show me where you got this bogus statistic from. My BS detector just went into overdrive.
  10. 7 billion, eh?[ Go to top ]

    Sorry, I got my statistics from Rolf. Remember, it's the trends that matter. ;-)

    Peace,

    Cameron Purdy
    Tangosol, Inc.
    Coherence: Easily share live data across a cluster!
  11. JBoss[ Go to top ]

    I think that Marc Fluery accounts for about 7.1 billion - 1 of those downloads sitting at home with his cable modem. I never understood these JBoss people talking about the number of downloads, when they download the product all the time. If so many people download it, why can they only list a few organizations actually using it in production?

    WebLogic, WebSphere and the lesser knowns like Oralce, Sun, Borland are real products with real customers. JBoss is just a lot of talk and an ego trip.
  12. JBoss download stats[ Go to top ]

    While the number of JBoss downloads is not a good statistic for determining the # of actual deployments in the real world, I think we all know the 7 billion is a mistake. Billion should obviously be million. Sourceforge has the stats for jboss publicly available and website itself states 2 million downloads last year.

    We should chalk up the 7 billion number to miscommunication and move on. The fact remains that JBoss is generating a lot of interest as an alternative to commercial offerings like Weblogic and WebSphere. Let's not get all tangled up about 7 billion.
  13. JBoss download stats[ Go to top ]

    While the number of JBoss downloads is not a good statistic for determining the # of actual deployments in the real world, I think we all know the 7 billion is a mistake.

    Mistake? No. It was a joke.

    I thought it would be funny to say it because every time the BEA vs. IBM market share thing comes up, someone pipes in and says "Well, what about JBoss, it had a zillion downloads!" So I made up a number and saved them the trouble. ;-)

    Regarding the story above about IBM throwing in licenses, that's how it works. IBM offered a company in our area a huge (many millions of dollars) discount on their CICS contract for them to buy an amount of WebSphere that was smaller than the CICS discount. In other words, they were paying them to take WebSphere (if you know the whole mainframe thing, these contracts are not considered optional.)

    In the end though, it doesn't matter. IBM has bought some real market share, and the product works (although it can be very painful at times, particularly if you don't use WSAD and do advanced stuff). It's being used in some really high end accounts, and it does seem that they are the definite #2 behind WebLogic in big J2EE datacenter applications. Maybe my view is skewed because we only work with high-scale (clustered) applications, but I'm guessing it's pretty indicative of the high end J2EE market. Secondly, we have a number of OEMs that sell J2EE "packaged" (or semi-packaged, i.e. semi-custom) software, and they almost all have to support WebSphere now, although originally they almost all supported only WebLogic.

    Regarding JBoss, there are definitely JBoss apps in deployment out there. They aren't necessarily using all the J2EE features (because JBoss hasn't traditionally performed well at all with advanced features like entity beans or JMS). However, I can think of at least a dozen of our customers that deploy on JBoss. Some of them chose JBoss because they needed very sizeable clusters and they couldn't afford to spend for commercial licenses for a whole bunch of servers, so it was worth the extra effort to get JBoss to do what they needed (and from their stories, there was some extra work, but they are happy now).

    The reason though that WebLogic and now WebSphere will still do OK even if 7.1 billion people download JBoss is because big companies (the ones that buy lots of licenses) don't like to support 100 different products. They want one computer vendor, one OS, one office suite, one mail program, etc. In the data center, it's a little different: they want easy-to-manage and never-go-down systems. Enterprise systems are never particularly easy-to-manage, so they don't want multiple types of enterprise systems in the same space (e.g. app servers), because they have to pay experts to manage them, so they will support one -- or at most two. Right now, IBM and BEA are the safe choices there. Occasionally Sun or Oracle. Even if JBoss some day were to catch up in performance and quality (this is just IMHO - I know some people will argue that it's already there), it will still have to become accepted in the datacenter as a manageable system. JBoss comes from a very effective direction ... through the developer and the cost-conscious manager. The result is that it will continue to get in at places willing to take the time to learn how to support it (kind of like Linux ten years ago.)

    Peace,

    Cameron Purdy
    Tangosol, Inc.
    Coherence: Easily share live data across a cluster!
  14. JBoss download stats[ Go to top ]


    Regarding the story above about IBM throwing in licenses, that's how it works. IBM offered a company in our area a huge (many millions of dollars) discount on their CICS contract for them to buy an amount of WebSphere that was smaller than the CICS discount. In other words, they were paying them to take WebSphere (if you know the whole mainframe thing, these contracts are not considered optional.)
    >>>>>

       I am currently working in a project where client received BEA Weblogic 6.1 licesne free as part of buying a group of ES 7000 hardware. Later the vendor tried to teach the client about J2EE technology and looted client money for last 4 years as part of consulting Fee. My client was doing business in last 100s of years in IBM mainframe. They decided move into web enabled services using BEA weblogic, JAM etc. Years ago started a number of J2EE projects without even having a single J2EE expert in their full time employee list. Consulting firm brought 100s of contractors from third world and took millions of dollars with BEA free license sales technology, so I wanna let you know that all of these guys are very smart enough to do "Buy one Get another one free, but give me an option to serve you" kind business approach.
       Story is not ending there, >>>> client was waiting for the application delivery from the vendors, after a couple of years, there is no result, later they got fucked up and turned into IBM( their old mainframe vendor) to finish the J2EE project started with free BEA weblogic and Weblogic JAM license. Now IBM is executing the project and using BEA application server. This was the best example of idiotic business approach BEA and their business associates follows to sell hardware and software. The client was not even 1% ready to move out of Mainframe based technology, they didn't have enough technical resources and they believed in BEA and their business partners, got into buy one get one deal and got fucked off.
      So let us stop talking too much good about BEA. If you guys have some kind of business deal with them, you can do that outside serverside.com
  15. TQ: JBoss download stats[ Go to top ]

    TQ, May be using your real name gives some credibility to your experiences?
  16. TQ: JBoss download stats[ Go to top ]

    TQ, May be using your real name gives some credibility to your experiences?


    Isnt is enough to know that he used to work for BEA and didnt like the experience?
  17. TQ: JBoss download stats[ Go to top ]

    |Isnt is enough to know that he used to work for BEA and didnt like the
    |experience?

    I have a strong suspicion that BEA didnt like the experience either. :)
  18. BEA[ Go to top ]

    TQ: So let us stop talking too much good about BEA. If you guys have some kind of business deal with them, you can do that outside serverside.com

    My comments were my own opinion, and as objective as one's own opinion can be.

    To be honest, despite a long history of being involved with WebLogic, it was IBM that sponsored us as a partner first, and IBM that first started sending us customers first. So, while I still have a lot of constructive criticism of WebSphere the product, I do really appreciate how well we've been treated by IBM, and I think that other companies could learn a lot from how IBM focuses on growing their marketshare in strategic areas. You might perceive that I'm bashing IBM for giving CICS discounts to grow their perceived WebSphere marketshare, but you'd be dead wrong -- it's a good strategy if it works, because CICS (etc.) is not the future of IBM, it's just a good lever that they happen to hold from the past. However, just because it's good strategy doesn't excuse people in the industry from understanding what it is, just like people should understand that JBoss download numbers don't equate 1:1 to production deployments in Fortune 100 datacenters.

    Peace,

    Cameron Purdy
    Tangosol, Inc.
    Coherence: Easily share live data across a cluster!
  19. BEA[ Go to top ]

    Cameron's got some good points, and I want to clarify my points earlier:

    I do think that IBM is using good strategy to get Websphere in the marketplace, and I don't have anything against their strategy. Whether we/you like it or not, marketing is important. We all rag on marketers, but I took some marketing courses in my masters program and learned a great deal about how to approach customers, work with them, and how marketing provides dollars for the techs to do more and better stuff.

    Whether or not you believe it, from a marketing perspective, you techies don't provide dollars, the marketers do. You may have the product, but the marketers get it to the right people, for the right price, and keep them coming back (if they're any good). They know how to develop relationships (customer-brand, customer-company, customer-product, customer-customer), loyalty, brand, and yes hype.

    IBM's got good marketing and good business strategy.

    My sole point is that, since these studies tend to be based on revenue in some way, and the groups like Gartner just take the numbers that IBM/BEA etc gives them, then we need more information. We don't really know the revenue/cost-center splits for the WebSphere product line, or the others as they are tied in.

    That doesn't mean that the numbers aren't correct. It just means that their are different ways of getting the numbers, or even looking at them. Should it be revenue based? Should it be license based? Or number of installs? Whatever. I just think the methodology should be made quite clear so that people can know how the Gartners/Giga's arrive at their numbers. Then we can make our own judgments. Knowing the final numbers without the underlying stuff makes the final numbers less valuable.

    Just saying that a company has XX% market share doesn't really mean much (except to the marketers who are maybe doing a good job here too). Anyone who's ever taken a stats class knows that this is useless without some key info like variance and std dev. But it's more useless without knowing the underlying methodologies.

    Jason McKerr
    Northwest Alliance for Computational Science and Engineering
  20. Just saying that a company has XX% market share doesn't really mean much (except to the marketers who are maybe doing a good job here too). Anyone who's ever taken a stats class knows that this is useless without some key info like variance and std dev. But it's more useless without knowing the underlying methodologies.
  21. BEA story[ Go to top ]

    <t q>
    This was the best example of idiotic business approach BEA and their business associates follows to sell hardware and software.
    </t q>

    Let me make sure I understood your story correctly. A company with zero J2EE expertise got free WebLogic licenses and, predictably, ended up not being able to deliver the product. Then IBM consultants came in and finished the project for them.

    Bottom line: it's BEA's fault.

    Fascinating logic.

    --
    Cedric
    http://beust.com/weblog
  22. JBoss download stats[ Go to top ]

    While the number of JBoss downloads is not a good statistic for determining the # of actual deployments in the real world, I think we all know the 7 billion is a mistake.

    >
    > Mistake? No. It was a joke.
    >

    The only thing that's really been proven in this thread is that some people don't recognize a joke when it slaps them in the face. Thanks for the laugh Cameron, and all those who responded to it.

    I wonder what the world population CGI script shows us up to now? ;-)

    onya - boonie.
  23. The mathematics of a joke[ Go to top ]

    There's probably some kind of formula g=x*10<super>3y</super>, where x is within an order of magnitude of the actual amount, and g is the number of gullible readers still buying the premise. We already know that for y=1, g=20 or so. So assuming a standard bell curve, set y=2 if you still want to possibly have fun with one or two people. Set y=3 if you really want to make sure you boink everyone over the head with a ball-peen hammer. And voila... 7.1 quadrillion.

    -Matt


    > The only thing that's really been proven in this thread is that some people
    > don't recognize a joke when it slaps them in the face. Thanks for the laugh
    > Cameron, and all those who responded to it.
  24. Keeps growing...[ Go to top ]

    Seen this...been there.

    Dont we have this discussion every year? :)

    BEA zealous fans keep complaining and WebSphere keeps growing...life goes on.

    Take it easy, Guys.
  25. Keeps growing...[ Go to top ]

    These are WEblogic fans. BEA haven't done much into Weblogic other than buying that product and included some Mbeans and colors of screen plus some web applications. The problem with BEA is, later with the money got from .COM market, they went and bought some more products, but this time things are completely different, they got a collection of garbage such as portal, workshop, rocket jvm etc. Then they thought about go and challenge everybody like Sun, IBM, and Oracle etc. They got that power from the .com age arrogant people joined with BEA through deals of portal, workshop, rocket engine etc. Now they are barking on the streets of Information technology, they talk too much loud against industry proven companies such as IBM, SUN, Oracle and some time even to Microsoft. Whenever I hear about their street barks, i just think about the story "Barking dog never brave enough to bite".
  26. The truth is out![ Go to top ]

    BEA haven't done much into Weblogic other than buying that product

    >and included some Mbeans and colors of screen plus some web applications.

    I just compared my vintage 1997 Tengah edition with WebLogic 8.1 and I must say, you're absolutely right, TQ: only the colors have changed! Outrageously, the WebLogic download is now much bigger, possibly due to their switching to larger fonts in the docs. Well, either that or Cedric has been inserting millions of bogus comment lines in the EJB container source code and tricked the compiler into leaving them in.

    I confess I downloaded WebLogic 16 million times myself in the past 7 years - only Cameron bettered me on this. So you know what their download statistics are worth!
  27. Anybody wanted to know why I am laughing on comparison between BEA and IBM, let us try to explain in some technical point of views.

    I am going to have a simple look into some of the BEA products and its internals implementations ideas in a developer’s point of view. I am not the guy who wrote the Java code behind their product, so I do not have complete pictures. But those who use them can easily identify some of the funny things.

    BEA’s portal architecture approach.
        Each portal has a number of pages, Each page have a number of portlets. Each portal,pages,portlets etc have implementation classes, each portlets have a xml meta data, in addition to that all of these informations persisted into a database. Looks good right??. Build and deploy portal application by half an hour!!!! After paying $250,000 license fee, wow wonderful idea.
        As a developer, just apply an SQL monitor tool to collect the database operation executed to bring up the front page of your portal. If you are a real developer, you may get a heart attack, because that number is too much and never able to explain in a resource oriented software developer’s point of view

        Not stopping there, try to look into some code BEA developers wrote for portal implementation, there is a custom debugging feature, each line you see some code like the one below
        If(debug.ON){
            Debug.out(“””” blah blahs ete);
    }
       If you ever look into the implementation details about the debug feature, you will come to know that cuz of the Senior developer had some clay in his brain, each programmers who used his debug implementation code had to use “If( debug,ON) 1000s of time without any reason. There was easy way to set up the "ON" and "OFF" features into each debug implementation classes, so that developers can just write some like
        debug.out( "remove the clay from BEA developer's brain");
    instead of write
        if(debug.ON){
          debug.log("God save these idiots from their darkness so that Java will be saved");
    }
       
      They do have a cache mechanism in BEA portal, I just wonder if they really smart and good enough to question me and all other developers in this world, how come they didn’t make that cache implementation cluster enabled and applied in their portal running of their own application server???
         I know the reason, they cannot fool around the customer and tell them their cache or cluster is fast enough to carry all the data they put into their cache in portal implementation

    Anybody has used BEA’s JAM??
     They talk too much about their JAM and tell the customer, hey we have a distributed transaction manager, we have the tools to generate the Java codes within seconds from COBOL copybooks. Well let us try to do some business using JAM.
    Create a java class from a copybook and try to analysis the code by applying some metrics gathering features such as
    Cyclomatic complexity, Efferent Coupling, Lack of Cohesion, Number of fields, number of methods, number of Levels, number of parameters and weighted methods in java file, !!!!!!!!!!!!!!!!. You will get a result where you can easily see that these auto generated java code is going to heat the CPU and JVM

    Finally, Wanna bug on BEA’s class loading design???
    Write an application with an assumption that, separate package for utilities classes, exceptions classes, system level factory classes in application level. Apply preferWebInfClasses craps into BEA web application deployment framework. Cuz of the confusions among class leader in BEA WLS , WEB application will crash within seconds!!!/
       
       Finally I have noticed in portal application, they have implemented a custom made Mbean to manage a web application resources( portal) such as behaviour traching, cache management ete from WLS console through a web application which report to the console. I asked them how come I can associate a Mbean with their WLS server so that I do have some additional features in WLS console. You know what was their answer, "it is not of your business!!"

    Some kids in serverside.com will be too happy by just seeing the console color changes in each BEA version. I am not happy, that is why I laugh when somebody compare BEA with other companies.

    Finally, kids, good luck with your enterprise dreams
  28. Regarding the debugging, that seems to be an available feature for log4j as well. Something like

    if (logger.isErrorTrue()) {
      logger.debug("");
    }

    It seems pretty standard. You don't HAVE to use it, but you can, and it probably speeds things up slightly. I'd do the same if I were writing a container.

    Why are you talking about implementation issues of your former employer? That's not very professional. Who cares what they do? They don't let you muck with JMX to the level that JBoss lets you do. That's not news.

    I had issues with Weblogic's classloading, too, but you're full of sour grapes.
    Steve

  29. They don't let you muck with JMX to the level that JBoss lets you do. That's not news.
    >>>>
       I know exactly how to muck into their JMX. In my current weblogic application, I do have features where WLS administrator can flush the cached data from Weblogic console. so what?. It is just a java application, where any kid can play around, if he know what is he doing.
  30. Of course you do. That's why you had to call support.
  31. I called their support cuz my client has already paid for product support, also some time I would like to check the support team memebers knowledge in J2EE and their own product. I still remember the chineese guy from New Jersey office. He is wonderfull to confuse and re-define J2EE technologies, so some time it is funny to call such people for time pass. Got it?. Still another reason to Laugh on BEA :))
  32. Regarding the debugging, that seems to be an available feature for log4j as well. Something like

    >
    > if (logger.isErrorTrue()) {
    >   logger.debug("");
    > }
    >
    > It seems pretty standard. You don't HAVE to use it, but you can, and it probably speeds things up slightly. I'd do the same if I were writing a container.

    I agree. In fact, I would say it's the only proper way to do it. Suppose, for example, that I want to log some complicated debug information, such as the contents of a row in a database. Using T Q's strategy, you would write:

    debug.out("Here's the row: " + fetchRowFromDataBase());

    But now the row is fetched from the database, even if debugging is OFF. The proper way to do it is:

    if (debug.ON) {
        debug.log("Here's the row: " + fetchRowFromDataBase());
    }

    which is remarkably similar to what is happening inside WebLogic according to T Q. So T Q, who's got clay in his brains here?
  33. The truth is out! and I am laughing on BEA[ Go to top ]

    Hello Pato Loco
           I have to call you "Idiot". I am sorry you are still working in IT, May be you have worked in a number of .Com s. Anyway , whatever down here is to just clear the clay from your brain and if still u have doubts, please try to look for a job in McDonalds.
         What ever the I have mentioned in my previous post is nothin but an implementation of factory pattern. I have a class called Debug on which I have a method called getInstance(). I have a debug,properties file where I have entries for each class I need to debug as given below.
    Myclass=remote
    Yourclass=log
    BEAclaybrainClass=log

    and also in my application where I have some other classes which I do not wanna debug right now .
    Irrespective of whether I have an entry for each classes in Debug property file of my application. I have the following type of code in my each application classes.

    Debug debug = Debug.getInstance(className.class);
    When ever my application load my class, they get a debug instance on which I debug my class.

    Depends upon the valus I have specified for each classes in Debug property file, I get a debug instance, which is nothin but a debugImpl classes. If not entry found in property file I get default impl class called silentDebugImpl class where Debug.ON = false;

    And all other implementation classes irrespective of whether it is for a Local DebugImpl or remoteLogImpl you get an instace of Debug class where Debug.ON = true.
      So simply by removing the entry from the debug.properties you can enable or disable the debug feature on a class instance running or used under your application. Or changing debug properties value from remote to log or local , u can change the way application logging the information same time at different places.
      So either u get a silentDebugImpl instance where no debuging feature enabled or you will get a Debug instance where implementation is different according to the corresponding value in your debug properties.
      So why the heck a programmer to check as given below
      if(debug.ON){
         debug.log("remove the clay");
      }
      
      by just putting
       debug.log("remove the clay from his head to save java");
    according to the implementation type of above debug instance , it will print the string.
      if the above debug instance is an instance of SilentDebugImpl, it won't do anything cuz ON value is false on it and that value control the debug class print statement
      if the above debug instance is an instance of RemoteDebugImpl or LogDebugImpl, the class will write into corresponding media, cuz the controlling boolean value is set true in both cases.
      You can control the boolean value in impl classes, Checking that value again in program which use the debug features, programmers just adding additional codes without any use. Also ON is a final variable which set as either true or false during the time constructor of impl class called which eventually call the constructor of super class.
      So just change the property file to enable or disable or change the debug mode, nothin else.
      Got it??/

    End of discussion
  34. fields vs. methods[ Go to top ]

    debug.ON is a field. If it is a constant field, the compiler will emit different code for whether it is set to true or false. It is a legitimate construct for certain types of conditional inclusion of debug output.

    Peace,

    Cameron Purdy
    Tangosol, Inc.
    Coherence: Easily share live data across a cluster!
  35. fields vs. methods[ Go to top ]

    ========================
    public abstract class Debug {
    private static final Debug SILENT_DEBUG_INSTANCE = new SilentDebugImpl ();
    private static Debug CONSOLE_DEBUG_INSTANCE = null;
    private static Debug LOGGER_DEBUG_INSTANCE = null;
    private static Debug REMOTE_DEBUG_INSTANCE = null;
    private static final boolean USE_PACKAGE_NAMES =
    DebugUtil.getPropertyAsBoolean("usePackageNames", false);
    private static final HashMap instanceMap = new HashMap();
    public final boolean ON;

    protected Debug(boolean isOn) {
    ON = isOn;
    }
    ========================================
    Not static, just "public final" which controlled by the implementation class's constructors.
  36. T Q is so full of it ...[ Go to top ]

    Why don't you post your full name here? This anonymous postings aren't doing your credibility any good (not that you had any to start with). Whoa man, get a life and stop paying your "girlfriends" by the hour! :-)
  37. T Q is so full of it ...[ Go to top ]

    Hey Cary,
        Any week end plan???
    I don't have a name man, it is all gone since I stopped working in BEA :)).
    Now my only job is sit here, work in Eclipse,Ant and J2EE and shut them out and let them know that, this is 2003 and their share value is single digit and there is no more .com s left and living with a J2EE container is not a full picture of enterprise computing.
  38. T Q sez "any weekend plan"

    Er?

    But honestly, your comments are pitiable and reek of a person who is desperate to draw attention to himself. A few hours with a shrink might do you some good and also get rid of all that megalomania!

    Have a nice weekend and remember to pay your GF's by the hour! :-)
  39. The truth is out! and I am laughing on BEA[ Go to top ]

    If debug.ON evaluates to false, you still have wasted time on creating the String (your debug message). Checking log.isDebugOn() is there for performance reasons.

    Elementary, T Q.
  40. The truth is out! and I am laughing on BEA[ Go to top ]

    There are plenty of check on Impl class such as

       public final void log(Object obj) {
    if(super.ON){
    logger.debug(DebugUtil.formatOut(StringUtils.toString(obj),1));
    }
       }
    As long as if(super.ON) false , there is not even a single process on "obj".
    But if you wanna argue for CPU level processing or JVM level processing or Single CPU vs Multiple CPU level, Hey I don't have time cuz micro level techology is still in its childhood. My major was Electronics and Communication where, we used to study transistors, diods and capacitors, so let us stop going into micro level.
  41. The truth is out! and I am laughing on BEA[ Go to top ]

      So why the heck a programmer to check as given below

    >   if(debug.ON){
    >      debug.log("remove the clay");
    >   }
    >   
    >   by just putting
    >    debug.log("remove the clay from his head to save java");

    Because if the debug message is a bit more complicated then this, for example, debug.log("Here's the record: " + fetchRecordFromDataBase()), then the very expensive method fetchRecordFromDataBase() that should only be called when debug is on is also executed when debug is OFF. If you put the debug.ON test around it (which is very cheap) this doesn't happen. Do you see my point?
  42. The truth is out! and I am laughing on BEA[ Go to top ]

    I am not writing Strings from bible into debug classes, i am sending some small bytes of stirng to track my code. Anyway, I wanna see my java class without much ifffffffs, it looks ugly and as long as I have another level of check in impl class such as
     public final void log(Object obj) {
    if(super.ON){
    logger.debug(DebugUtil.formatOut_(StringUtils.toString(obj), 1));
    }
    }
     As far as objects passed to the debug instance, they have already created as part of programs execution such as result from a private method or an exception. So as long as I am not trying to send stories from Bible to debug instance and I am not doing any string process before executing the real method in Impl classes. I still save time, create good java code, and get same performance.
      Got it??/
  43. T.Q = Trolf Throllerud??? Hum.....
  44. I am not writing Strings from bible into debug classes, i am sending some small bytes of stirng to track my code.

    ....
    >   Got it??/

    Not really. Just because you do that doesn't mean that other people don't have more CPU expensive debug methods. Debug methods allow you to debug, either in development or later on in production, problems that are hard to track down.
    It is not uncommon that these call diagnostic methods that can take time to do their work, and if you multiply that time by the total of all the methods that have debug, it will have an effect on the running time of the program.

    if (Debug.ON)
    {
       Debug.log("log some info",
                 callDiagnosticMethodThatTakes500ms(),
                 callAnotherMethodThatTakesTime());
    }

    Experience, and running Quantify, tells us that the calls to log() and to the diagnostic methods DO affect the overall execution time of the program.

    And as Cameron points out, if you use the Debug.ON construct the compiler will optimise your code away and your program will have a smaller footprint.

    Paul Parsons
    The Server People
  45. TQ no matter who you are, but I think its time to close the discussion.
    I read the whole threads. I worked with WebSphere, with WebLogic and with JBoss in production environments, for differents kind of app. I am not working for a software company. Just acting as a consultant like you try do.

    I thanks BEA for having kicked you out - from what I read - if its damn true.

    You said BEA is making business from a J2EE container. It seems to change in 2003 isn't it ? Have you ever heard about fortune's company needing complete IT infrastructure, not just J2EE, like business process management, portal solution, data integration, JMS middleware ?

    You illustrated with a log implementation that you state from beeing from BEA.
    As quoted by others, it is a common log syntax pattern. So common it is documented in the FAQ of Apache Log4J as a performance optimization. May be you never get there ?

    You changed your mind by saying it was not log(string) but log (object). Wich is a different assumption. If as a regular user I write log("it is " + a + b + c) this is different from log(a), isn't it ? I see here 3 toString() call and 4 string concatenations without StringBuffer used in the first log(string) before the method log(string) itself just get called. Are you assuming how user will use your log "framework" ? What if this user is the software engineer of the company you work for who has to take the project when you leave ?
    You could say at least "ok guys, you are right, I understand what was behind now"

    Have you ever thought that "if (Debug.ON)" could be added by a source preprocessing tool as a transparent optimization pattern in the build process ?

    Are you making so much noise for such a simple things as "log for debugging internal purpose only" ? Damned, BEA products line will reach far beyond hopefully without your ideas and the time we loose answering you.

    Alex
  46. Well,
       Now your pick is just Logging mechanism and trying to prove their bad design is just for internal purpose. Also you and some of your friends here seems like too much attached with string operations. Anyway in my understanding BEA designed the Debug functions with a lot of other operations in mind, not just printing your string only to the log file.
       Second, Debug issue is just one thing I have mentioned about BEA products. I have also mentioned a number of issues such as
     1) BEA Weblogic Server class loading problems
     2) Issues of using preferWebInf in an enterprise IT environment and breaking of web applications.
     3) BEA JMX implementations and easy way to hack it to add additional management functionalities into weblogic console, some thing like they itself done in web logic portal applications(cache, behavior tracking etc).
     4) BEA Portal design issues
     5) BEA portal application's database dependency to display each page in applications
     6) Bad design on generating java code in BEA's JAM tool Egen and applying metrics tools to evaluate the generated codes.

     In my previous project, IT manager asked me to show him below 3 sec response from their portal applications, decided to build on BEA portal. He had already a serving portal build in house using Microsoft technologies, which serve with 3 seconds response time under small loads. They decided to move into BEA to handle the heavy load and also get the good side of Java into their applications. Do you know one thing, but just porting the in-house portal application into BEA portal framework, it takes more than 5 seconds to show up the page cuz of BEA's fucking design to just build a simple customized web page and show to the browser. I asked the manager whether he had any kind of assurance from BEA sales people about the response time of the portal application. He said no, so who is the bad guy here, is it BEA who sell some bad products with their industry name or a developer who has to use it. Anyway we had 100s of way to still keep BEA portal in place and get a 3 sec response time by looking into the application and make necessary changes as we needed.

       You have the full right to use BEA application as you like. I have the full right to point out the troubles in their products and ingnore their attitude as they are #1 in Middleware technologies.
  47. This is getting way off topic, but T Q's anonymous trolling has made me think of some knowledge that is useful to pass on to the general community:

    This also may explain the "if (Debug.On)" pattern that T Q did not like seeing in java source code: At BEA, the "if (Debug.On) {}" pattern is commonly used to improve performance, to "hide" internal-use-only code, and to reduce .class file size. If "Debug.On" is a static final set to false, then javac is smart enough to totally skip creating and putting the debug byte code into the generated .class files. If on the other hand, the "Debug.On" were moved into the debug method, the call to the debug method itself would not get compiled out - lowering performance and allowing programmers to decompile the shipped .class files and see the debugging code.

    T Q brought up complexity analysis tools, which, as one would naturally expect, BEA engineers will sometimes use on internal code. I take complexity analysis tools with a grain of salt, as they can base their calculations on faulty initial assumptions that can throw their results way off. And, of course, they can not have any knowledge of the semantic complexity of the code beneath. For example, in many of these tools, each "case" within a switch statement is considered a branch, so a large switch statement immediately causes a jump in complexity measures - but would one really want to remove case statements in favor of "less complex" multiple if statements?

    Tom Barnes, BEA, Liberty Corner, NJ

  48. T Q brought up complexity analysis tools, which, as one would naturally expect, BEA engineers will sometimes use on internal code. I take complexity analysis tools with a grain of salt, as they can base their calculations on faulty initial assumptions that can throw their results way off. And, of course, they can not have any knowledge of the semantic complexity of the code beneath. For example, in many of these tools, each "case" within a switch statement is considered a branch, so a large switch statement immediately causes a jump in complexity measures - but would one really want to remove case statements in favor of "less complex" multiple if statements?
    >>>>>

       That means in order to create a regular Java bean class for an entity with get and set methods of all its attribute, We have to increase the complexity into the level of 99999999!!!!!. Good idea.
    You know what is wrong with your EGen??. If not I know, you guys applied same logic to convert all types of COBOL Copybooks without considering the size and complexity of individual copy books, down the road result is irrespective of whether it is a simple copy book or complicated one, generated java code is nothing but a complicated one and telling the JVM to have a long ride to process individual attributes of a java object. Dude, I think JAM was belong to some one else, just bought it , changed the colors and plugged into the BEA without looking what it is cooking inside. That is the truth
  49. Numbers?[ Go to top ]

    So you're saying that Gartner didn't do their research here, and really has no way of knowing relative market shares? But you do? I'm not saying I know for a fact either way (read the article, but haven't seen the paper), but Gartner does this sort of thing for a living. I have a hard time believing they're just pulling these numbers out of no where.

    Can you provide statistics and research to back up any of your statements?

    --Kevin

    Discliamer: I'm not affiliated with either company, and use WebLogic for my own work.
  50. Numbers?[ Go to top ]

    I have downloaded Jboss more than 5 times... but I only use WebSphere and Weblogic...
  51. GARTNER IS EVIL[ Go to top ]

    Gartner says BEA is not number one, therefore gartner is evil.

    I should know I was trained in the UK as a computer professional and I can tell when gartner is being evil.

    I know they are lies, damn lies and reports from gartner.

    evil
  52. re: Numbers?[ Go to top ]

    Ofcourse the research firm do publish their number based on some facts. But there is a always a catch. They are more like Financial Analysts;).
    "IBM gained market "share of new-license revenue" at the expense of BEA and others"

    Is it overall numbers of only new license?

    As said, IBM may need to break out their numbers. MQSeries now WebSphereMQ is also Websphere, but does not count as Java Application Server..
  53. Numbers?[ Go to top ]

    Kevin,


    The Gartner reports tend to be revenue/licensing based, but the revenue figures come directly from the customer. I spoke with them once and asked them about this, and it was confirmed by them. So right off the bat there are some anomolies. Weblogic licensing revenue is probably direct and easy to create statistics for since they don't have a lot of tie-ins.

    However, we don't know how IBM accounts for licensing/revenue in pure accounting terms, and what they report as licensing or revenues for the Websphere product. Maybe you buy a big-dog computer from IBM, and they give you a free websphere license to use it. Does that count as licensing (through the revenue model)? What if the owner of the licesne doesn't use it?

    What if IBM convinces them to use the product through tie-in. "We'll give you a discount if you also use DB2, or our IDE?" How is the revenue then reported in terms of both accounting and licensing? Are they using revenue basis similar to activity based accounting? What percentage of the transaction is attibuted to which product? Or is it just licensing?

    This is what also makes me skeptical of IBM's report that they've made back all of their invesment on Linux? How do they define this revenue streams/ROI analysis? Did they just make the money back? Because that would a bunch of bs. It has to be and ROI or economic-alternative analysis...

    The Gartner I spoke with said they don't really do all that much research. They take the data that the vendors send them and churn out reports based on it. Given IBM's product tie-ins and the fact that we don't know how they account for licensing/revenue-to-product, it does seem that the research is flawed. They're not really *doing* research. Their doing analysis on supplied numbers, and there a zillion different ways that companies come up with these supplied numbers.

    I don't think the burden of proof should be on the person asking for more details from Gartner, I think it should be on Gartner to back up their numbers, how they arrived at them, and how IBM and BEA arrived at theirs.

    Jason McKerr
    Northwest Alliance for Computational Science and Engineering
  54. Numbers?[ Go to top ]

    oops. That first line shoud say

    "The Gartner reports tend to be revenue/licensing based, but the revenue figures come directly from the *VENDOR*."

    Jason
  55. True Enough, But ...[ Go to top ]

    I can agree on most of what you said here. My problem is that statements contradicting this have the same burden of proof. If we think that Gartner didn't do their research well and therefore their conclusions are suspect, we have to turn that around on someone that hasn't even done that much investigation--polled people or checked the numbers (or anything else as far as we know)--and states that the Gartner conclusion is incorrect--that person has no basis for conclusion. At best, that person can admit no evidence either way.

    --Kevin
  56. True Enough, But ...[ Go to top ]

    I can agree on most of what you said here. My problem is that statements

    > contradicting this have the same burden of proof. If we think that Gartner
    > didn't do their research well and therefore their conclusions are suspect,
    > we have to turn that around on someone that hasn't even done that much
    > investigation--polled people or checked the numbers (or anything else as
    > far as we know)--and states that the Gartner conclusion is incorrect--that
    > person has no basis for conclusion. At best, that person can admit no
    > evidence either way.

    I agree that you can't *prove* their research is bogus without evidence, but they're the ones who are claiming that they have discovered something, and that it's newsworthy. That puts the burden of proof on their side. Just as I would be justified in dismissing someone who claims the Earth is pear-shaped based on research methods they refuse to detail, without having to do my own research to prove that they're wrong.

    My experience has been that Gartner and most other "industry analyst" research houses produce, among other things, reports that are bought and paid for by a vendor to cast their product in a favorable light or make it look more successful than it is, and industry predictions that sound authoritative by assigning precise probabilities to trends that no one has a way of measuring with any accuracy. I've often thought I could be quite successful as an industry analyst, if only I had no shame...

    So, no, I can't prove they're wrong. But based on past experience, I can be justified in ignoring them, and recommending that others do the same, unless they provide any credible evidence to support their pronouncements.

    -- Jim
  57. Can you trust anyone?[ Go to top ]

    Jim,

    If you have problem with Gartner and its report, then wait for Giga, IDC studies on the app server market and integration market. If you have no faith in any of these analysts, then period.
  58. Internal Logic[ Go to top ]

    Jason's critique is one of methodolgy. The fundamentals of IBM's and BEA's licensing conventions are known. The methodologies of analysis are manifold.
     If you agree with Jason's critique of Gartner's methodology then you agree that Gartner has a flawed process no matter what data Gartner uses.
     If you disagree with Jason's critique of Gartner's methodology, you should be able to defend Gartner's method of analysis.
  59. Another hoax?[ Go to top ]

    http://www.sys-con.com/java/readerschoice2003/liveupdate.cfm?BType=4


    Regards,
    Horia
  60. Gartner in general[ Go to top ]

    Now I wont say Gartner is totally useless since Sr Management love their meetings etc however I have been in several Gartner Teleconferences to include one in which they stated OS/2 would be the dominate OS of the 90's. Oh well one wrong but that was a damn big one.

    I really dont believe they are objective nor do I think that its possible for anyone to really get objective stats when the vendors may have a lot to lose by not being completely honest. We all know statistics can be made to say about anything.
  61. <snip>
    Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone. ;-)
    </snip>
    Well, how many projects were deployed on JBoss in a production environment out of those billion downloads last year?
    And how many just downloaded for keeps ?

    IMO, these app servers numbers does not mean anything and it does not reflect quality of the product.
    We all do better service to the J2EE community by ignoring these reports.

    Move on folks... There is nothing here..
  62. <snip>
    Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone. ;-)
    </snip>

    There must be a serious defect in JBOSS Download site. It seems that it is multiplying the actual download numbers by a very large number. They better fix or no one will trust their download numbers.
  63. 7.1 Billion served[ Go to top ]

    <snip>

    > Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone. ;-)
    > </snip>
    >
    > There must be a serious defect in JBOSS Download site. It seems that it is multiplying the actual download numbers by a very large number. They better fix or no one will trust their download numbers.

    I'm pretty sure Cameron was joking with the 7.1 billion figure.

    Whatever process Gartner uses to calculate marketshare (at the application server level, product suite and other categories), up until now it had WebLogic ahead and now it has Websphere leading. So unless they changed their process, Websphere seems to be growing compared to others.
    Clearly the growth is not from the typical readers of this site, but that does not mean it is not happening.
  64. 7.1 Billion served[ Go to top ]

    John, keep in mind that IBM has recently completed a few significant acquisitions (Rational, for instance) and many of Rational's products are being renamed WebSphere something or the other. So even if Gartner maintained their old methodology of counting (which is to ask the vendors), the fact that IBM now has a few dozen additional products labelled "WebSphere ..." would definitely tilt the scales, don't you think?
  65. |
    |Clearly the growth is not from the typical readers of this site, but that does
    |not mean it is not happening.
    |

    Its growth Jim, but not as we know it.

    I can tell you where at least some of this revenue comes from. We can extrapolate from there what representation of reality these figures are.

    A company I know is a large client of IBM's.
    In with the purchases of mainframe hardware, Lotus Notes software, etc etc are a very large number of Websphere Appserver licenses (100's) are thrown into the deal.
    Do they use Websphere? Yes, but not in the numbers they actually pay for. Essentially they get a better deal on the other stuff if they agree buy some Websphere as well.

    So, if you hazard a guess at how many other big IBM customers get the same kind of treatment (and I know they do), then its not too difficult to see how IBM have pulled off this remarkable turnaround.

    -Nick
  66. websphere= bowlingball in pipeline[ Go to top ]

    Clearly the growth is not from the typical readers of this site, but that does not mean it is not happening.


    EXACTLY!

    All along IBM has been the marketing department for BEA. What I mean is that many big blue customers start out with websphere but as they ramp up they run into problems and then move to WebLogic. see :
    http://www.bea.com/framework.jsp?CNT=pr00860.htm&FP=/content/news_events/press_releases/2002

    So if this news represents a "baby boom" (big "IF") of websphere adoption among laggard RPG and Cobol developers, then its reasonable to expect that the marketsthare ratio will return to historical norms once these customers mature, become educated and eventually migrate.

    Matt
  67. All along IBM has been the marketing department for BEA. What I mean is that many big blue customers start out with websphere but as they ramp up they run into problems and then move to WebLogic. see :

    > http://www.bea.com/framework.jsp?CNT=pr00860.htm&FP=/content/news_events/press_releases/2002
    >
    > So if this news represents a "baby boom" (big "IF") of websphere adoption among laggard RPG and Cobol developers, then its reasonable to expect that the marketsthare ratio will return to historical norms once these customers mature, become educated and eventually migrate.
    >
    > Matt

    Matt,
    In the interest of full disclosure, I'll state up front that I work for IBM.
    I have to say though, that a press release from BEA, about BEA, hardly proves that they are winning battles against IBM. You can dispute exact numbers, but by all measurements (except those from BEA), WebSphere is gaining marketshare.

    You've certainly put a unique spin on it however. The BEA sales force must be quite excited about the upcoming spike in sales.
    I guess time will tell what those "laggard RPG and COBOL developers" do.

    If you have an affiliation with BEA (or other vendor) please disclose it.

    Tom
  68. Mitchell Brothers ...[ Go to top ]

    Tom, appreciate your full disclosure re: working for IBM. Would you also be able to do likewise re: how IBM accounts for WebSphere sales? Is it just the app server or the whole kit and caboodle?

    Inquiring minds want to know!
  69. WebSphere Sales[ Go to top ]

    Tom, appreciate your full disclosure re: working for IBM. Would you also be able to do likewise re: how IBM accounts for WebSphere sales? Is it just the app server or the whole kit and caboodle?

    >
    > Inquiring minds want to know!

    The numbers are only publicly reported at the brand level (all WebSphere products ).
    By the way, Rational products have not been renamed WebSphere (as you stated in another post) and are not included in WebSphere revenue.

    With regard to other peoples comments that IBM is "buying marketshare", I can only say I wish it were that easy. Every WebSphere sale I know of was won in a highly competitive environment. In almost all cases, usability, manageability and performance studies were done, and the competition discounted their software as much as IBM (except for JBoss :))
    This is what J2EE is all about. Competition based on features and TCO. If IBM was able to buy marketshare, I wish we would have done it sooner....

    Tom
    The opinions expressed are mine, not IBMs.
  70. IBM buys marketshare[ Go to top ]

    With regard to other peoples comments that IBM is "buying marketshare", I can > only say I wish it were that easy. Every WebSphere sale I know of was won

    > in a highly competitive environment.

    Tom, with all due respect I find this *extremely* hard to believe. Numerous postings here have alluded to the fact that IBM virtually gives away WebSphere. I know you work for IBM and hence have to no option but to stick to the party line, but the truth of the matter is quite different. IMO, you are better off not posting such blatant mistruths which will only serve to erode your credibility!
  71. WebSphere Sales[ Go to top ]

    The appserver is only one peice of the puzzle. For any non trivial application you end up buying Network dispatchers, load balancers, webservers, databases, security software, LDAP, Message Q, operating systems etc and then a lot of hardware to host all these. Apart from all that, one needs to buy support for all this.

    IBM is trying to sell a total solution all under one umberella with one support aggreement. So now providing a one stop shopping suddenly is a bad thing because BEA does not provide all this.
  72. WebSphere Sales[ Go to top ]

    Rao, huh?

    The question here is what does IBM mean by "WebSphere" sales and for Gartner to then use this number to say that WebSphere has overtaken WebLogic! If WebSphere means 500 different things and WebLogic is construed to be just the appserver, common sense dictates that the comparison is fatally flawed.
  73. WebSphere Sales, ( J2EE technology sale)[ Go to top ]


    The question here is what does IBM mean by "WebSphere" sales and for Gartner to then use this number to say that WebSphere has overtaken WebLogic! If WebSphere means 500 different things and WebLogic is construed to be just the appserver, common sense dictates that the comparison is fatally flawed.
    >>>>>
      Again, Andy, you sounds like a fresh college graduate, trying to cover an enterprise application development proceess either by reading some books or just playing with an application server. How long you need to go back to school to understand, that just buying an application server won't bring anything into a client's IT division other than just a Java application called J2EE container????

    Regarding whether I got fired from BEA.
      It is none of your business. I know exactly what I am doing in my job, so I am not a person depend on employer to pay my bills. By the way just got a call from a firm in Michigan, telling me that BEA consultants are colleting huge money from a client who don't have much idea about J2EE and Weblogic. They wanna get a good Weblogic & J2EE guy in office. Are you out of job or hanging there for your employer's blessing???. If you are really smart go and get that job.
  74. T Q!!![ Go to top ]

    Hey TQ, I know you from your postings in the "Sun outsources some work to India" thread. Weren't you the guy who said you have a dozen girlfriends or something like that? (without mentioning the fact that you had to pay them by the hour ...). No wonder you got fired from BEA, couldn't concentrate on your work, eh?

    I see no reason for you to get so defensive when asked about your past life (unless there's something to hide, of course).

    T Q > /dev/bull!
  75. WebSphere Sales[ Go to top ]

    |
    |IBM is trying to sell a total solution all under one umberella with one support
    |aggreement. So now providing a one stop shopping suddenly is a bad thing because
    |BEA does not provide all this.
    |

    No, not at all. There is no problem with this. Its quite a good thing SO LONG as buying one product doesnt mean you are up for buying the whole shop to get anything to work. (ie the Microsoft Way).

    Most companies have existing (non-IBM) investment in all the infrastructure you just mentioned and they are not going to throw that away just to bring in an appserver.

    I think where IBM have been traditionally criticised, has been in the area of getting their products to work with non-ibm products. Either its not possible, or it requires a batallion of IBM consultants to get it to work.

    The whole BEA vs IBM thing aside, the thing you can say about BEA has been that they have always had a vested interest in being as easy to integrate as possible. They support the largest number of JVM's, hardware, databases, etc than most other vendors. This is largely because
    a) this sells
    b) they dont have other "levers" (like hardware sales) to pull.

    I personally am always wary of the "software stack" vendor. They have a vested interest in selling you the whole stack (whether you want it or not).

    -Nick
  76. BEA Press doesn't prove anything[ Go to top ]

    I think serverside.com had already published 100s of comparison between BEA and IBM, but still I just wonder how come somebody can compare these two company. BEA looks like a small division of company like IBM,SUN,Oracle or Microsoft. I have worked with BEA and I was out there with BEA sales people. In my personal experience, BEA sales people are nothin but a group of junks at the same time IBM people are smart and well trained to make business from any client. Even after the great disaster of .com time, some people write here with a sense that an application server is the complete picture of an enterprise application and I hope that thoughts leading them to compare BEA with IBM. IBM have a complete business solutions in place which include Hardware,Software and a well established professional service team, at the same time BEA laid of their 70% of PS and remaining working with the support team. A couple of years ago BEA could able to charge the license with a heavy price and could able to be a leader in application server market, but in current situation , customers are looking for a complete solution such from a vendor who can provide all kind of service from same source. IBM , Oracle, Sun and Microsoft is far far better than BEA in a customer's point of view.
       Those who don't have enough money, they can play with J2EE and down loaded version of Weblogic or some open source app servers or can try to build from open source group of products.
      Those who has enough fund they are going behind a stable company whom they can trust for their future business. BEA has to go a long way to reach that position. This is my personal opinion after working with that company for a good time.
  77. Gartner: IBM buys marketshare[ Go to top ]

    TQ, your comments are so laughable! Why the bitterness? Were you fired from BEA for incompetence, perhaps?
  78. BEA Press doesn't prove anything[ Go to top ]

    I agree with you that sometimes I get tired with the endless post on theserverside about who is ahead in the application server market. Always very headline grabing stuff with little substance.

    I did find interesting is what you said about the sales experience. We had completely different sales experience two years ago. IBM sales did not look very impressive when dealing with a green field developement where as BEA sales was a lot more on the ball. In the end my company picked BEA Weblogic.

    As for PS I was not very impressed with BEA and found their architecture team lacking. Also they did not deal with a mixed team of their and our people very well. Lot's of times things turned into battles between the BEA way and a more open technology path.

    In terms of integration we integrate BEA weblogic with Siebel, JDE Edwards and various Corba systems. We also integrate with most of the main retailer chains like Wal-Mart, Target, Walgreens etc. We just announced our 500k subscriber.

    David
  79. JBOSS IS EVIL[ Go to top ]

    JBoss dominates the market and comes from the US, therefore it is EVIL

    I don't believe they ever had 7.1 billion downloads it is a lie.

    Lies are EVIL
  80. Tom,
    I am not asking anyone to take my word for gospel. The anecdotal evidence contained in the referenced document ("more than 210 customers in the Americas, Asia/Pacific and Europe selected BEA over IBM in head-to-head sales situations during the past quarter. In 125 of these cases, BEA was selected despite the fact that IBM WebSphere was the incumbent platform. This includes 94 instances in which WebSphere customers selected BEA for new projects, and 31 scenarios in which IBM WebSphere was removed and replaced with BEA WebLogic™. ") speaks for itself.

    My point was just that statistics can easily be misinterpreted (lies, damned lies, etc.) and that in light of the more reliable anecdotal evidence, IBM is clearly not becoming more competitive.

    Another case in point is Dell replacing Websphere with WebLogic (less than 1 year ago):
    http://www.bea.com/framework.jsp?CNT=pr00797.htm&FP=/content/news_events/press_releases/2002

    Matt
  81. The title says it all. Anecdotes and quotes that come straight from the BEA website or from BEA executives at a press event (as yours have) are more than likely biased in favor of BEA. It's hard to see how they would be "more reliable" than a report from Gartner.

    My favorite quote on this story came from the Associated Press wire service: "BEA officials did not return calls seeking comment."

    Randy Schnier
    I don't speak for IBM
  82. read my original post[ Go to top ]

    Before I started having to defend against personal attacks, I described how additional datapoints provide some insight into a reasonable explanation that takes all the data into consideration...

    I haven't seen any data -except TQ's post ;-)) - to challenge this explanation. Do you have any?

    Matt
  83. Before I started having to defend against personal attacks, I described how additional datapoints provide some insight into a reasonable explanation that takes all the data into consideration...

     
    > Matt

    Matt,
    I don't see any personal attacks on you in this thread. Even TQs comments were not directed at you.
    I do not see any datapoints from you either. The PR from BEA is not a datapoint, it is a marketing ploy. IBM did make some comments in response, pointing out the obvious misinformation (http://www.eweek.com/article2/0,3959,476252,00.asp), but a response wasn't even needed. Customers saw it for what it was.
    Are you really an unbiased observer or do you work for BEA? That answer would help put your comments into the proper perspective.

    Tom
    The opinions expressed are mine, not IBM's.
  84. Ask Dell or IBM![ Go to top ]

    Look Tom, if it were untrue IBM would have a stronger rebuttal case -like a Dell correction...

    and you wouldn't have Analysts accepting it as true:
     "Mike Gilpin, an analyst with Giga, said he can see both companies having a valid point of view, "yet IBM still is growing revenue in this space while BEA is not. Yet we have to be careful in looking at IBM's revenue, because it's part of a larger whole and more difficult to analyze clearly, since it's not broken out separately in the full financial reports.""

    Needless to say, IBM can show increasing license revenue due to "umbrella accounting" and BEA can show license revenue reduction due to customers buying weblogic more through their portal, integration, and platform products. My point, supported by the 200+ customer examples, is simply that the new websphere license sales are destined for WebLogic eventually.

    have a nice weekend,
    Matt
  85. And then one would correctly assume that all of the people migrating from BEA to JBoss now, will eventually get the people to migrate from WebSphere as well. Regarding Garner, they are the biggest joke in the industry. All research is bought and paid for, everyone knows that.

    People on this board need to get a sense of humor and realize the 7.1 billion downloads were a joke.

    Only time will tell what happens, personally, I am guessing that IBM, BEA, and Sun are Gartners top three paying customers. Any way to find this out? Gartner is a joke as are most of the analysts. I thought their credibility was gone long ago, after they predicted every .com industry was the next multi-billion dollar industry.
  86. <j2ee master>
    And then one would correctly assume that all of the people migrating from BEA to JBoss now, will eventually get the people to migrate from WebSphere as well. Regarding Garner, they are the biggest joke in the industry. All research is bought and paid for, everyone knows that.

    People on this board need to get a sense of humor and realize the 7.1 billion downloads were a joke.

    Only time will tell what happens, personally, I am guessing that IBM, BEA, and Sun are Gartners top three paying customers. Any way to find this out? Gartner is a joke as are most of the analysts. I thought their credibility was gone long ago, after they predicted every .com industry was the next multi-billion dollar industry.
    </j2ee master>

    Gartner may be joke. But the people who decide which app server to buy are not technical all the times and they do not consider gartner as a joke.

  87. > Gartner may be joke. But the people who decide which app server to buy are not technical all the times and they do not consider gartner as a joke.

    That is why so many companies waste money and never get the results they should from technology. We as technologists need to get rid of companies like Gartner, they suck!
  88. 7.1E9 downloads[ Go to top ]

    <snip>
    Besides, they both have fewer downloads than JBoss, which logged 7.1 billion downloads in April alone. ;-)
    </snip>

    Maybe it should read "... 7.1 billion BYTES of downloads in April alone. "

    :)
    Boris
  89. Unless there is a way to independently verify the license numbers from IBM, we will have to take this with (more than) a pinch of salt! Which WebSphere product is IBM referring to? At last count, there were over 200+ products that were called "WebSphere ..."

    There are also reports that even IBM's earnings might not be what it appears to be! See recent article "Think Twice About IBM's Earnings" at:

    http://online.wsj.com/barrons/article_print/0,,SB105155090564619100,00.html

    Also, inquiring minds want to know: where does the Oracle app server rank? As is typical of Oracle, one hears a lot of bluster from them, but it is all "big hat, no cattle" as Scott McNealy would put it :-)
  90. JBOSS might have millions download per year. But it all use less. What do you do with a product when it does not give you the documents to start with.
    I know people will now point to me that they do have a free online document. But those who have tried to use it will find it frustrating.

    It can't be used even by college students for learning J2EE and EJB.

    And they still call themselves part of open source community. These are the people who will destroy open source community. People should not get mislead by them.
  91. I'm wondering how can they ask money for such a "marketing product" like websphere. Believe me, I've worked with both WS and Weblogic, but you can compare Weblogic with WebSphere. Even JBoss or OC4J it's much more better than WebSphere. With miles ...
    Regards.
  92. Just a developers point of view:

    We do provide server side components (www.servletsuite.com) and our stats is based on users who paid for commercial deployment, asked about some add-ons, reported bugs etc. (in other words not just downloaded out stuff):

    Weblogic is still on the top (and was there always, versions are started from 6.1)
    Interesting, that for the last 3-4 month JBoss related requests are going down (no idea why, maybe it is just a fluctuation, maybe not).
    The second position now is the trio Orion, JRun and probably Resin
    Sun ONE (of course, from this point of view only) does not exist at all :--) But in the same time Iplanet is still in the game.
    WebSphere (again, from this point of view only) is actually not far from Sun ONE

    Dmitry Namiot
    Coldbeans
  93. what you may not be seeing[ Go to top ]

    is the "enterprise", er, "corporate" market. Lots of big corporations like IBM. It makes them feel warm and fuzzy. And if they ever get cold feet, IBM has consultants to hold their hands to keep them warm. A lot of self-hosted corporate shops go with Websphere, because BEA is perceived as a "startup"
  94. As core functionality of app servers is becoming more of a commodity and focus is shifting towards addressing integration challenges (such as asynchronous conversation with Web services, coordination of flow and long-running transactions), how are IBM and BEA positioned to compete against each other with their respective platforms. Now that Sun (and who knows, maybe Oracle soon too) joined IBM and BEA in the BPEL TC camp (OASIS), isn't the platform war open again for grabs?

    Cheers.

    Jill.
  95. I wonder how much IBM paid for this "research"...
  96. <cnet_article>
    Sun ONE (Open Net Environment) and Sybase's Enterprise Application Server run a distant third and fourth, with 4 percent and 2 percent market share, respectively. Other, smaller companies fill out the market picture.
    </cnet_article>

    This would mean that Oracle 9iAS has less than 2% of the market and is among "smaller companies"... Can this be the case?
  97. There has to be an ism or slick buzzword that sums up the scenario when products win on technical merit but fail as a total package. WebLogic and JBoss win on technical merit but can't compete against Big Blue on marketing, channels, etc... Hell, JBoss is giving it away. Look at Microsoft vs. Apple.

    I'd love to see TSS come up with an ecomometric formula explaining buying decisions. decision=(.8)business case + (.2)technical case.
  98. There has to be an ism or slick buzzword that sums up the scenario when products win on technical merit but fail as a total package. WebLogic and JBoss win on technical merit but can't compete against Big Blue on marketing, channels, etc... Hell, JBoss is giving it away. Look at Microsoft vs. Apple.

    >
    > I'd love to see TSS come up with an ecomometric formula explaining buying decisions. decision=(.8)business case + (.2)technical case.

    You would actually find that the technical case coefficient (tcc) is industry and application dependent. With larger banks, telcos, insurance, logistics etc requiring much more technical capability than smaller retail, manufacturing, or distribution businesses.

    You would also find that for many applications and environments the tcc rises at a pretty fast rate.

    Therefore, marketing, channels, etc. may provide advantages initially but over time, tcc becomes more important and technical merit becomes a long-term advantage.
    Matt
  99. percentage of what?[ Go to top ]

    IBM at 37%, BEA at 29%..

    OK, but in Juni the whole "Big App Server market" will be only 50% of what it was the year before..

    Regards
    Rolf Tollerud
  100. I wiork with App serwer for 3 years and my experience lets me say tahtb WebLogic is fairly more advanced and powerfull product taht WebSphere. It is sad true that companies like Microsoft or IBM are so strong in marketing that tahey are able fire anybody with better solutions.

    sad man
  101. The article clearly says that this is an App Server base. Later, he goes to say that another seperate Portal marketshare. If IBM just used the WebSphere name on everything, then we would be number one in all the races. These numbers a clearly the App Server and WebSphere products that run on top of the App Server Only. BEA also includes Portal on top of WebLogic a App Server sell. This number does not include DB2, Tivoli, Rational, or Lotus all seperate divisions within the software group. Nor any WebSphere named product that does not run on top of the App Server.

    Most people here I bet just tried like WebSphere 3.x and still use that as a basis comparing it to the latest version of WLS or JBoss (even though this discussion should not include this, I have to because of all the JBoss people the chime in with the majic down load). I guess that makes Eclipse (built and still developed by IBM developers) number one in the IDE market.

    So I guess I can say WLS stinks because I tried version 4 or JBoss is immature because I tried version 2. If you haven't used WAS version 5.0, then your comments are unfair.

    Notice I did not bash current versions of any products. WebSPhere 5 is a strong solid product. And I work for IBM but I have used WLS 6.1, 7, 8.1 beta and I have used JBoss 3 and 4. Honestly, these are all strong products and have an advantage somewhere or another. I have seen bad configurations and code from admins and coders and then blame the app server. Good architects and programmers can make good stuff work on any platform.

    Comments my own and not IBM
  102. I was working with BEA WebLogic, before we have shifted to WebSphere 5.0. In order to shift to WebSphere, we had to make a report on the technical evaluation report on WebSphere and also case studies of various companies using it. In my evaluation report, surely WebLogic was the winner in terms of usage, performance, maintenance. WebSphere was far far behind all the above, but the compnay(the managers) still chose WebSphere. Why?? Because WebSphere is cheap and the kind of contracts they make with our company (in terms of partnerships and I don't know what else).
    Now we are using WebSphere 5.0 and it is stuffed with problems. Since 1 yr. we are trying to fit WebSphere into our environment and we are facing lots of problems. One being the rigid structure of WebSphere installation package (the MQ Series part always installs to a fixed directory location).
    Now IBM has release fixpack1, which has many conflicts with VisiBroker version. The reasons we found was that IBM has placed JDK 1.4 features with JDK in fixpack1 (rt.jar file). This is not acceptable.
    From the experience with IBM, my personal understanding is that, they do not have a broader view of the type of usage of their product, or in other sense, their design is quite poor. Mostly they look for quick and dirty solutions, instead of a better design solution.
    I think the reason for IBM being #1 is because of its kind of marketing and the current market conditions.

    Anyway, we are married to WebSphere and cannot divorce it in the near future.

    Cheers
  103. Considering that a Resin setup blows all competition, why can't set up a price for the worst system, Weblogic or Websphere?

    I might be willing to contribute a small sum..

    Regards
    Rolf Tollerud