TMC Releases J2EE Appserver Marketshare Research Report

Discussions

News: TMC Releases J2EE Appserver Marketshare Research Report

  1. TMC has released the results of a Sun-sponsored J2EE Application Server research report that measures market share by use in production and hands-on experience. The results paint a very different picture of the industry with BEA in a comfortable lead over IBM in production, Tomcat/Jetty leading in development, and Sun (including J2EE RI) leading in hands-on experience.

    You can download the survey from TMC's middlewareresearch.com here:
    http://www.middlewareresearch.com/endeavors/040827J2EESURVEY/endeavor.jsp

    Interesting results include:

    • Sun and its many brands (including the J2EE RI) scored the highest share by hands-on experience.
    • BEA is first in production use with a strong lead over IBM; Tomcat/Jetty almost tied IBM for 2nd place.
    • JBoss came in 4th in both production and development use by product.
    • Performance, Stability, and standards conformance were the top 3 selection criteria. Interestingly, support/training/consulting services came in 10th.
    • If you include non-J2EE options, Tomcat/Jetty leads by a by a comfortable margin in terms of development use.
    • Corporate Architecture Mandates was the primary reason a particular application server was selected.
    The report also includes other statistics such as database use, vendor trust, types of projects the respondents were working on, and more.

    Threaded Messages (30)

  2. IPlanet, Sun ONE, Sun Java System Application Server are all the same product and J2EE SDK has nothing to do in this context. So the Sun score is totally fake. Aside from that I notice that Tomcat/Jetty took first place as usual.

    That BEA Weblogic and IBM Websphere together have only 30% of the market in development and 36% of the market in production certainly will be a surprise to some. That development is lower than production also suggest that their market share is slipping.

    Regards
    Rolf Tollerud
  3. Very interesting report if you can read in it that people are using
    'Apache geronimo' in a production environment. I would like to know what kind of production environment this is seen the current status of this project?
  4. That development is lower than production also suggest that their market share is slipping.
    I don't think that it suggests that at all. It suggests that people do not pay for licenses for development servers. They use tomcat instead. It also suggests that J2EE apps are portable enough that you can develop on one and deploy on another. If this suggests anything at all as far as a trends in the industry it is that apps are not using things like EJB that Tomcat/Jetty do not support.

    Michael
  5. It's good to be the Sponsor[ Go to top ]

    Sun sponsored the survey, and strangely enough they were the beneficiary of the results that most brazenly threw math out the window.

    In the Marketshare by Company, TMC aggregated positive responses from products like IPlanet and SunOne, which are simply successive versions of the same product.

    Why not add up all the responses to people who have used Weblogic 7.0, 8.0, and 9.0?

    You could have made the equally arbitrary decision to sum up the uses of JBoss and Tomcat, and JBoss Group would have been the clear winner among vendors.
  6. It's good to be the Sponsor[ Go to top ]

    people who have used Weblogic 7.0, 8.0, and 9.0?
    Or 7.0, 8.0, and 8.1. Whatever.
  7. In one case[ Go to top ]

    I agree that the Sun products should not have been roled up into one for "Mindshare" since a lot of people who responsed for the Sun products most likely used more than one of the Sun products (since they are really the same product offering) and checked off multiple Sun products.

    But in the "Market Share" section they SHOULD be rolled up into one since they are just different versions of the same product line. If they were not rolled up in the "Market Share" section then you would also have to separate out WebLogic v7, v8, etc. into separate sections.

    I don't know why J2EE SDK is in their and I'm not sure what it is since it is not an application server?
  8. Aggregated and non aggregated[ Go to top ]

    Hello,

    Just a note from TMC:

    Some people here had questions about the aggregation of the brands of J2EE Application Server by company. The Application Servers were shown two different ways--by individual product and by company. The data is the same for both charts. If you prefer to view the chart by product, it's broken out that way for you. If you prefer to view it by company, it's set out that way. Sun Microsystems as the sponsor was interested to know if the brand fragmentation was impacting the perception of thier products, and indeed having multiple different names for a product seems to have had that effect.

    There could be other artifacts of the way the question was presented, controls for presentation of the question were not in the scope of this study.

    Definitely on occasion there are some "interesting" responses such as (sic) "using Geronimo in production" or "Not using Eclipse because of price". We expect to see some number of such responses, and find that they are not a statistically significant portion of the overall sample.

    We do try to comment in the study on the sources of bias in the study that we think could be statistically significant, as you can see if you download and read it. Survey research has a human element which makes it closer to social science than traditional lab science, and so of course the results should be taken as a statistical aggregate. Still, we feel these results are interesting, and accurate to the extent possible.

    Miko Matsumura
    Industry Analyst
    The Middleware Company
  9. Aggregated and non aggregated[ Go to top ]

    Sun Microsystems as the sponsor was interested to know if the brand fragmentation was impacting the perception of thier products, and indeed having multiple different names for a product seems to have had that effect.
    So if BEA was the sponsor, they probably could have gotten you to add up people who have used Weblogic Application Server, and the Weblogic Platform Server to determine BEA's marketshare. And to, um... understand the impact of brand fragmentation, of course.

    I do appreciate that your 'Disclosures' section was much better written this time. But the one-sided way you choose to do company-level aggregation was a transparent attempt to reward a sponsor, who otherwise would not have had much to crow about in this report.
  10. Aggregated and non aggregated[ Go to top ]

    The BEA servers ARE also aggregated! Same for IBM ,Oracle etc. If you do not ask people whether they use Weblogic 7 or Weblogic 8 but simply ask whether they use weblogic you will count all versions.

    If you ask people whether they use the SunONE application Server or the iPlanet Application server, they will only indicate 1, not all and so each instance is counted once and there is no harm done at all in aggregating this. Matter of fact, if Sun as a sponsor had chosen to list their entry as 'a Sun appserver' we wouldn't have this discussion.

    I never was a big fan of Sun, have used mainly BEA in projects until recently. Now, as a freelancer working for a customer with limited $$, we decided to use the Sun Java System appserver platform edition 8 - and it is excellent. For production we actually moved to the Sun Java System Appserver 7.1 Standard Edition- and it is super, for a price that can't be beaten. I seriously looked at JBOss as well, and they will get a hard time. Sun's appserver is simply (a lot) better and either free or cheap, depending on the scalability/loadbalancing/HA features you need.
  11. J2EE SDK should be included in the mind share count because it includes an application server. For example, J2EE 1.4 SDK includes Sun Java System Application Server Platform Edition 8.

    Paul Ko
    Sun Microsystems
  12. It's good to be the Sponsor[ Go to top ]

    Sun sponsored the survey, and strangely enough they were the beneficiary of the results that most brazenly threw math out the window.In the Marketshare by Company, TMC aggregated positive responses from products like IPlanet and SunOne, which are simply successive versions of the same product.Why not add up all the responses to people who have used Weblogic 7.0, 8.0, and 9.0?You could have made the equally arbitrary decision to sum up the uses of JBoss and Tomcat, and JBoss Group would have been the clear winner among vendors.
    The only problem with this suggestion is that Apache Tomcat doesn't come from JBoss Inc. It comes from Apache.

    Like many other commercial products and open source projects, the JBoss Server uses Apache Tomcat, Apache Axis, Apache Logging, Apache Jakarta BCEL, Apache Xalan, as well as Apache Jakarta commons projects...

    - geir
  13. It's good to be the Sponsor[ Go to top ]

    Sun sponsored the survey, and strangely enough they were the beneficiary of the results that most brazenly threw math out the window.In the Marketshare by Company, TMC aggregated positive responses from products like IPlanet and SunOne, which are simply successive versions of the same product.Why not add up all the responses to people who have used Weblogic 7.0, 8.0, and 9.0?You could have made the equally arbitrary decision to sum up the uses of JBoss and Tomcat, and JBoss Group would have been the clear winner among vendors.
    The only problem with this suggestion is that Apache Tomcat doesn't come from JBoss Inc. It comes from Apache.Like many other commercial products and open source projects, the JBoss Server uses Apache Tomcat, Apache Axis, Apache Logging, Apache Jakarta BCEL, Apache Xalan, as well as Apache Jakarta commons projects...- geir
    JBoss Inc. funds a very significant portion of Tomcat development efforts.

    Bill
  14. It's good to be the Sponsor[ Go to top ]

    Sun sponsored the survey, and strangely enough they were the beneficiary of the results that most brazenly threw math out the window.In the Marketshare by Company, TMC aggregated positive responses from products like IPlanet and SunOne, which are simply successive versions of the same product.Why not add up all the responses to people who have used Weblogic 7.0, 8.0, and 9.0?You could have made the equally arbitrary decision to sum up the uses of JBoss and Tomcat, and JBoss Group would have been the clear winner among vendors.
    The only problem with this suggestion is that Apache Tomcat doesn't come from JBoss Inc. It comes from Apache.Like many other commercial products and open source projects, the JBoss Server uses Apache Tomcat, Apache Axis, Apache Logging, Apache Jakarta BCEL, Apache Xalan, as well as Apache Jakarta commons projects...- geir
    JBoss Inc. funds a very significant portion of Tomcat development efforts.Bill
    Apache Tomcat has a long history of commercial support of it's developers, and it's good to see it continue. This model of companies supporting developers, even if just part time, on open source projects as a way of giving back to the community is something to be encouraged, especially when these companies products depend so heavily on such a wide range of open source projects.

    geir
  15. It's good to be the Sponsor[ Go to top ]

    JBoss Inc. funds a very significant portion of Tomcat development efforts.Bill
    Which is great. But it's still an Apache project.
  16. Tomcat and corporate contributions[ Go to top ]

    Sun sponsored the survey, and strangely enough they were the beneficiary of the results that most brazenly threw math out the window.In the Marketshare by Company, TMC aggregated positive responses from products like IPlanet and SunOne, which are simply successive versions of the same product.Why not add up all the responses to people who have used Weblogic 7.0, 8.0, and 9.0?You could have made the equally arbitrary decision to sum up the uses of JBoss and Tomcat, and JBoss Group would have been the clear winner among vendors.
    The only problem with this suggestion is that Apache Tomcat doesn't come from JBoss Inc. It comes from Apache.Like many other commercial products and open source projects, the JBoss Server uses Apache Tomcat, Apache Axis, Apache Logging, Apache Jakarta BCEL, Apache Xalan, as well as Apache Jakarta commons projects...- geir
    JBoss Inc. funds a very significant portion of Tomcat development efforts.Bill
    In addition to what Geir said, I'd add that JBoss employs Remy, a former Sun employee who worked for years on Tomcat while working at Sun. Sun also employs four other people who continue to work on Tomcat while working at Sun. Other companies also fund people to work on Tomcat. None of this makes it anything other than an Apache project.
  17. J2EE = EJB[ Go to top ]

    When you say "non-J2EE" are you realy saying "non-EJB"?
    Geronimo has market share in production? That does not make sense.

    Are there large public sites that are using EJB and willing to say they are?
    It be an interesting question to see: JDBC vs Hibrenate vs iBatis.
    (ex: http://struts.1up.com has 10MM members and uses iBatis v2)

    You can see Eclipse dominate, and Tomcat "leads by a comfortable margin", which makes sense.

    .V
  18. Bias?[ Go to top ]

    I think following is more reasonable.

    1. Weblogic/Tomcat/Websphere deployed most in corporates, JBoss climb up very fast, deployed from software vendor to corporate now.
    2. JBoss/Tomcat as development platform among software vendors;
       Weblogic/Websphere as development platform as well in corporates.
    3. I wonder SUN RI or SUN products still valid.
       There are long time no active update on SUN java.sun.com/j2ee/blueprints..etc
       java.net has few articles on J2EE as well.

       J2EE last two years are driven by community efforts for J2EE framework, Eclipse, application..etc, SUN drives J2SE now but lost the J2EE world.
  19. When to use which app. server?[ Go to top ]

    One might choose
    * tomcat/jetty when no need for EJB
    * websphere when you need to play safe
    * jboss when you are experimenting -or- deploy when your app is not mission critical
    * weblogic whenever if you can effort
    * pramati when you want to try new things
    * oracle when oracle sales people convince you that Oracle AS runs way faster with Oracle db
    J2EE last two years are driven by community efforts for J2EE framework, Eclipse, application..etc, SUN drives J2SE now but lost the J2EE world.
    I think this is very good for the future of J2EE and maybe we should be thankful to Sun for letting us drive J2EE.
  20. Java Java system Server Rocks ![ Go to top ]

    I think following is more reasonable.1. Weblogic/Tomcat/Websphere deployed most in corporates, JBoss climb up very fast, deployed from software vendor to corporate now.2. JBoss/Tomcat as development platform among software vendors;   Weblogic/Websphere as development platform as well in corporates.3. I wonder SUN RI or SUN products still valid.   There are long time no active update on SUN java.sun.com/j2ee/blueprints..etc   java.net has few articles on J2EE as well.   J2EE last two years are driven by community efforts for J2EE framework, Eclipse, application..etc, SUN drives J2SE now but lost the J2EE world.
    Having used Weblogic, Tomcat, JBoss over the period of past 6 years, I think am eligible to say the following

    1. BEA Weblogic, Fantastic product in terms of usage and performance in stand alone to clustered environments

    2. Apache Tomcat, Excellent choice of servlet container, although Caucho Resin also looks good but requires some licensing issue.

    3. JBoss, Good & Bad, even though the memory footprint is small for this server and the performance is good, I will never advice anybody to use this bcoz of lack of documentation, lack of configuration environment/tools and being left helpless wondered like a lost kid in the woods in case of problems.Hmmmm 4.0 has a appplet embedded in the JSP called as admin console :-))

    4. Finally I found nirvana, when I went here http://java.sun.com/j2ee/1.4/download.html . Yes indeed the Sun Java system aPplication server is a ray of for common folks like me to enterprise managers. This piece of software cannot be gauged unless you open the admin console. The first look reminded of my old days of using the best UI admin consoles from Macromedia ColdFusion 6.0.

    I will say nothing in favour of SJSAS, instead you folks try it once, over your spare time during lunch hours and see how this piece of nicely organised software charms your :-)))

    VIJAY
    J2EE Architect
  21. RE: JBoss, Good & Bad[ Go to top ]

    - It should be advise (verb) not advice (noun) in
    "I will never advice ..."
    - JBoss documentation is available, I'm surprised that you don't know that.
    - JBoss support can be obtained for a fraction of money paid for Weblogic.
    - There is Web console and JMX console, not as elaborated as in Weblogic, though, but it comes for no money.

    So JBoss vs. Weblogic is like Unix/Linux with configuration files and vi vs. Windows with wizards and dialog boxes. If you don't know Unix then it is easier on Windows - click, click, click ... very cool.
  22. Update on Java BluePrints[ Go to top ]

    I think following is more reasonable.1. Weblogic/Tomcat/Websphere deployed most in corporates, JBoss climb up very fast, deployed from software vendor to corporate now.2. JBoss/Tomcat as development platform among software vendors;   Weblogic/Websphere as development platform as well in corporates.3. I wonder SUN RI or SUN products still valid.   There are long time no active update on SUN java.sun.com/j2ee/blueprints..etc   java.net has few articles on J2EE as well.   J2EE last two years are driven by community efforts for J2EE framework, Eclipse, application..etc, SUN drives J2SE now but lost the J2EE world.
    Actually the BluePrints has some great new content

    * Conventions
         Project Conventions, first draft at
    http://java.sun.com/blueprints/code/projectconventions.html
         Naming conventions, updated draft at
    http://java.sun.com/blueprints/code/namingconventions.html

    * Chat on Java BluePrints for SOA at
    http://java.sun.com/developer/community/chat/index.html
    on September 14. 11:00 A.M. PDT/6:00 P.M. GMT

    * Final version of BluePrints book online at
    http://java.sun.com/blueprints/guidelines/designing_webservices/

    * BluePrints Neighborhood on java.net at http://blueprints.dev.java.net/
    which is where the BluePrints team will do their development now

    * an updated version of the Java Adventure Builder, also developed on
    java.net at http://adventurebuilder.dev.java.net/

    * a new project, the Java BluePrints Solutions Catalog, which is
    available in early access since it is being developed on java.net at
    http://bpcatalog.dev.java.net/
    This will include periodic updates as this new project evolves.
  23. J2EE = EJB[ Go to top ]

    Are there large public sites that are using EJB and willing to say they are?
    Most of the financial institutions I've worked for are using EJB's i.e. session beans deployed to a container. Although they all seem to do it in different ways and with different architectures.

    Ian Purton
    Website and Server Monitoring | Email Marketing | Website Analysis
  24. Where's Orion AS?[ Go to top ]

    Where is Orion AS? That beats 'em all.
  25. Where's Orion AS?[ Go to top ]

    orion has become oracle's j2ee server...
  26. Great Survey, some intresting questions which I don't see in the results that would be nice to see are: "Place a mark next to the operating system you are using..." in Development and in Production. And "Which of the following service oriented architecture concepts have you used / will use within your organization?"
  27. Sorry guys, but this research report is almost totally meaningless. First, all the participants are self-selected. Actually, they're self-selected twice: once by being members of the opt-in mailing list and second by choosing to respond to the survey invitation. Anyone involved in serious market research knows that such self-selection is a big no-no, and generally means the result cannot be extrapolated beyond the sample group.

    Second, there's no clear definition of what the terms "development" and "production" mean, so we have no way to know what the respondents meant when they answered these questions.

    Finally, if I have a team of six developers who are using Tomcat on their workstations for development and all deploying to a single WebLogic server, does this mean that Tomcat is used six times as much as WebLogic? and therefore has six times as much "market share"? Obviously, that's a ridiculous notion.
  28. Thanks for your feedback[ Go to top ]

    Thanks for your feedback. TMC involves the community in both the production and analysis of the research. Part of that analysis includes critiques such as yours, and we encourage "active" rather than passive consumption of our research.

    Self selection produces bias when it is nonrandom, and we attempt to document the nonrandom elements of selection in section 4.7 of the report, titled "Selection Bias". All surveys short of a census have sample bias, which must be taken into account if generalizing the results.

    In terms of definitions, we try to define areas which we feel would produce significant ambiguity. At this point, we feel that the ambiguity implicit in "development" and "production" would not affect the results in a statistically significant way. In general, question ambiguity must be evaluated by critical consumers of survey research. This is why we include the original survey questions as presented, so that our readers can make their own determinations, as you have.
     
    Finally, if you have a team of six developers using Tomcat to develop, and each of them deploys on Weblogic, please consider what would happen if all six developers took our research survey. Each of them would respond that they used Tomcat for development once, and each of them would respond that they used Weblogic for deployment once, resulting in parity for both systems.

    In the big picture, we advise that consumers of all research evaluate the limitations and biases implicit in any study. We attempt to document these and foster discussion which are both part of our process for creating meaningful research.
     
    Thanks for your comments,
    Miko Matsumura
    The Middleware Company
  29. I just love TSS posters[ Go to top ]

    It always amuses me, no matter how bloated IBM's server gets, or how many employees BEA sheds, posters at TSS will back them to the last.

    Even if Sun produces a better app server they will blindly bash them in the hope that everyone will use WebLogic.
    For some actual facts visit.
    http://java.sun.com/developer/community/chat/JavaLive/2004/jl0427.html

    They seem to have forgotten where Java came from, and want to back also-rans like BEA.
  30. Cannot Afford Eclipse[ Go to top ]

    30 people in the survey want to use Eclipse but dont due to cost concerns.
  31. NetBeans too![ Go to top ]

    A number of people also mentioned cost as a reason they weren't using NetBeans!

    ...maybe something to do with thoe iPod prizes ;)...