Discussions

News: Cost benefits of using J2EE and Open Source

  1. Cost benefits of using J2EE and Open Source (112 messages)

    ZDNet has posted a brief article about the cost savings of using Open source J2EE solutions. The article profiles an insurance company project that used java and open source projects to integrate systems as diverse as IBM AS/400s, Windows 2000 servers and Oracle servers. Velocity, jboss, and eclipse were used in development.

    the article can be found at...
    http://techupdate.zdnet.com/techupdate/stories/main/0,14179,2908812,00.html

    Threaded Messages (112)

  2. Though I fully agree that it is perfectly possible to squeeze costs regarding development and deployment infrastructures while still remaining pretty productive, in reality, this only represents a small part of the decision-making process when a company select IT products. I'm fairly sceptic as to whether this is really going to change anytime soon.

    The architect of this company has made daring choices (for a corporate company) and has realised that the other side of IT software is not such a bad place to be, and on the contrary can be quite a pleasing experience. I'm pretty sure this sort of experience is not unique though. Actually, I believe that the more functionality is built atop application servers, the less expensive the licence fees will be. Quite naturally. Expensive costs will only move to another applicative layer. So the days of expensive licences for application servers are probably numbered, the days of expensive licences in general probably not.

    Overall, there is definitely an OSS momentum building as more and more proprietary products offer OSS "plug-ins". Portal, CMS, DM, CRM tools can be deployed on more and more application servers, of which JBoss. More and more IDEs offer Ant, Struts, TomCat, JBoss, ... (you name it) ... support. Still these proprietary products can cost up to hundreds of thousands of dollars in certain configurations... so it's really a cost shift. And this cost shrinking is usually only a small part of the total cost of ownership anyway.

    But as a developer, I certainly enjoy the wealth of open source initiatives Java offers in basically all fields of development, including database and integration. I also enjoy using some of these expensive proprietary tools too. OSS is not a magic bullet and OSS does not stand for automatic quality, but it widens the choice, which can only be good.

                    Yann
  3. OSS is here to stay.[ Go to top ]

    Interesting article, and I am sure there are hundreds of more stories just like this. OSS is a viable solution, and it is good to see more and more people adopting it. Currently, my company uses the following OSS projects in production:

    Log4j
    Xerces
    Xalan
    FOP
    Struts
    Commons
    Lucene
    POI
    ORO

    Along with this, we also use these in development:

    Ant
    Eclipse
    JUnit
    XDoclet (integrated as we speak)

    Granted, most of these are "support" libraries, but we have been please regardless. And in the situations where we have replaced a commercial solution with an OSS one (e.g. Lucene replacing Oracle search product) the results have been outstanding. We are still having a hard time pushing for OSS app servers/servlet engines, but we are trying (WebSphere rant excluded).

    A couple of points about the article itself...

    1. I would argue that cost savings is not even the biggest advantage of OSS. I would argue that there are two bigger reasons. One, in my experience, OSS is made by developers, for developers. It's easy to use, unlike many cumbersome commercial solutions. Two, it's OPEN. You can change it. You can debug it. You OWN it. If something is wrong, you can crack open the hood and fix it. You don't have to call a support line, open a ticket, wait, raise the severity of your ticket, wait, pay out the nose, etc.

    2. There was one comment that bothered me in this article:

    "This model worked well, but in order for us to use the full J2EE stack, we needed to add EJB and the other J2EE features that Tomcat was missing."

    This makes it sound like they wanted to use a full-blown app server so they could use EJB. I would argue that first, you have to have a NEED for EJB, THEN you make the switch. A slight, but important, difference. I think too many people want the whole J2EE stack when Tomcat may be a perfect solution. I am not saying that EJB don't have merit, just make decisions based on a technology need, not technology for technology's sake.

    My thoughts.

    Ryan
  4. OSS is here to stay.[ Go to top ]

    "Currently, my company uses the following OSS projects in production: Log4j, Xerces, Xalan, FOP, Struts, Commons, Lucene, POI, ORO. Along with this, we also use these in development: Ant, Eclipse, JUnit, XDoclet"

    I don't think there are many multiple-thousand-dollar products that Log4j, Xerces, Xalan, FOP, Struts, Commons, Ant, JUnit, XDoclet and ORO replace, so these cannot legitimally be called dary OSS choices. I think those OSS products that have a hard time jumping on the enterprise stage position themselves as direct competitors to more or less expensive solutions. Take Eclipse/JBuilder, JBoss/WebLogic, MySQL/Oracle, Linux/Windows, Apache/IIS, OpenSta-JMeter/LoadRunner, JetSpeed/SilverStream. These are the products that still undergo a lot of resistance for various arguably valid reasons.

    Also, on a personal note, as an architect, I would be a bit scared to recommend POI and Lucene in production for critical applications. They may be satisfying to fulfil some non-critical requirements though. But you cannot really compare things like Lucene/Intermedia and POI is pretty much a pretty good hack with a hard time to come in perspective of trying to keep up with the different MS Office formats.

                    Yann
  5. Why OS[ Go to top ]

    - Standard
    - Open in case you do not think it is secure or are suspicios.
    - Open in case you found a bug that must be fixed
    - Save time in budgeting meetings
    - Save time meeting with legal on license
    - Make money in production (Ex Tomcat/PostgreSQL can support 40,000 concurent users, BEA / ORACLE would take money)
    - Survivie economic downturn by keeping costs low (Eclipse, Struts)
    - Invest in people instead
    - It is more reliable
    - It is better supported

    Buying BEA/Oracle/Webshpere is like wearing Gucci shades..... you think you are cool.

    Why not OS?
    "If we give them lots of our money, they will support us better"
    I tried that with my X-wife.

    .V
  6. Why OS[ Go to top ]

    Don't get me wrong. I'm a happy open source software user and I'm quite familiar with the Jakarta suite, actually I use part of this OSSs on a daily basis. I'm also the first one to admit that in many cases, a full-blown application server is overkill and that yes you can do the same thing on TomCat and even faster. But it is a fact that customers are still scared of OSS for their critical applications. Reason #1 pertains to so-called professional support and continuity. Companies invest up to several millions euros/$ to get their applications running and they want guarantees that their product will still exist in a year from now. I agree that in reality, most software companies cannot truly guarantee that but it's all about risk assessment. OSS is still associated with too much perceived risk (even if in reality it does not hold so true).

    But OS does not mean standard no. And it also has its associated cost of ownership whether you want it or not. OS is not the way to survive an economic downturn: OS is only a free tool, it's how efficiently you use it that will keep your business going. OS is not more reliable and not better supported. These are all commonplace statements, which cannot be backed up. Some OSS is dead good, some other is plain crap, and guess what, OSS and proprietary software are architected/designed/implemented by people with roughly the same skillsets. People who write OSS often have a daytime job writing proprietary software.

    As I said, OSS offers a broader choice, that's all. And I'm quite happy to use any *good* software, be it OSS or not. It will take longer for customers to realise that JBoss is probably as good as its proprietary counterparts and it's a pity. Even longer for them to realise that JBoss is enough for 90% of the applications.

    Finally, I would never use a software because I think it's 'cool'. This is the fastest way to disaster. The unfortunate thing is that many use OSS because they think it is 'cool'. Just the attitude thing... anyway.

                    Yann
  7. Why OS[ Go to top ]

    "Tomcat/PostgreSQL can support 40,000 concurent users, BEA / ORACLE would take money"

    40.000 concurrent users ? What injector was used to measure such performance ? Is it 40.000 users with think time ? On what platform ? How many CPUs ? What sort of SQL queries ? Can you guarantee that it can hold that load for long without a crash or database corruption ? Can many other customers confirm that they successfully held that kind of load for long ? If yes, then I'll give it a try with a proof-of-concept project and I'll benchmark it against Resin/PostgreSQL for instance and then I'll take the winner and test it against Oracle with all kinds of SQL statements. Then, I'll be closer to my expectations.

    But not all applications need to support 40.000 concurrent users anyway. There are actually very few with such requirements. Usual requirements amount to 50-500 simultaneous users (no think time) and up to 500-5.000 concurrent users with think time. So TomCat or Jetty will do fine for most requirements. :)

                   Yann
  8. OSS vs. Commercial[ Go to top ]

    Well, I think some of these OSS products are giving their commercial counterparts a run for their money. Linux is making in-roads in the Windows server market. Apache is already the dominant web server. Eclipse is quickly becoming the choice of many Java developers.

    <tangent>
    I just don't get WHY anybody would buy JBuilder, with Eclipse and IDEA out there.
    </tangent>

    As for the BIG products (app servers, middleware, databases) commercial products still dominate. However, I promise you that there are many apps running BEA/IBM app servers that would work just as well running Tomcat.

    And about Lucene and POI - why are you scared to recommend them? We used Intermedia for a year. It was a huge PITA for many reasons. We switched to Lucene and have had no problems. As for POI, this was a specific circumstances. Our in house users have there data in Excel. The need to upload this data to our server, which is Java. We need to parse the Excel to extract their data. POI was/is an obvious, simple, stable solution. I am not suggesting rewriting MS Office on top of POI, but it you need to read MS documents in Java - it works great.

    Ryan
  9. OSS vs. Commercial[ Go to top ]

    Ryan,

    I've seen quite a few application-server-based applications that could run happily with a simple JSP engine. Actually, that sort of projects easily makes 50% of what I have seen so far.

    The reason why I would not recommend using POI is that it is the output of the reverse-engineering of MS Office formats. It probably works 90% of the time. Critical applications would require that it works 99.99% of the time. The difference is unacceptable.

    Now regarding Lucene, it is perfectly acceptable and pretty fast (at least to index and retrieve the text files on my hard drive). As I have not carried out any benchmarking against bigger text repositories, I cannot guarantee that it is scalable and that its performance holds. I cannot guarantee the contrary either. But, it does not support as many formats as Intermedia which comes with all necessary parsers. Then it is a niche market as it works best with file systems. Database systems almost always offer the same functionality natively. Also, it has tough commercial competitors. For critical text-indexing applications, I would think of Google search engine as an alternative. There are certainly cases where I would recommend Lucene (again probably more than 50% of the time) but that's mostly for non-critical applications.

                    Yann
  10. OSS vs. Commercial[ Go to top ]

    This has been a very interesting thread to read through on the opinions of open source. It seems as this post, hits it right on the head. About 50% of all applications out there need nothing more than Tomcat. I see this all the time. As the Director of Business Development for JBoss, I run across large commercial companies who have WASTED thousands of dollars on commercial application servers. If you analyze the application server industry, which I do all day long, one would find some very interesting conclusions. A lot of the money spent on app server technology was spent during boom times, when people did not care. JBoss and the OSS world are closing the gap at a rapid clip on the proprietary vendors. The app server industry like most industries lead by standards is commoditizing. We have gone from 30 vendors two years ago to maybe 6 today. The industry is revolting against high dollar software that will neve pay for itself. If everyone here worked for a company with a complete OSS development and deployment environment, would so many of your friends be out of work? Companies are realizing that it is more important to invest in people than empty promises of software vendors, OSS allows this to happen.

    We win against the big boys all the time these days and with companies that range all the way to the Fortune 10. These companies are not afraid to move the mission critical systems to JBoss. Why? Many would be suprised that cost was not the first answer almost 10 times out of 10. Acutally the software is more stable, with better support, and you can eliminate licensing constraints from your architectural decisions. If you were to add all of the app server's used in development and production since the J2EE app servers have been a going concern there would be less than 2,000,000 copies out there. JBoss was downloaded 2,000,000 times in 2002 alone. This enables extreme stability of the code base. As someone else mentioned earlier in OSS features are done for developers by developers. We do not try to sell people on features they do not need, like an integrated stack with all the bells and whistles, when all you really need is an app server.

    Follow the press, there will be a lot coming out about JBoss and some of the large companies who have made the move never to return to the world of commercial J2EE servers.

    Give it time, this is the year of OSS, unless we get back to boom times, which I suspect will never happen again in software. Just like the automobile industry of the 1920's, all mature industries consilidate. There will be IBM, Microsoft, and Open Source, we know that for sure. The others are all in a dog fight.

    Check out the press that will come out of Linux World next week, good things to come. Cheers and best to everyone.
  11. Jboss is hardly open source[ Go to top ]

    Ben, you've been drinking your own koolaid a wee bit too much.

    IMHO Jboss is about as open source as MS Windows. [Open source is not just about publishing your source code -- it is also about getting a body of independent developers to understand,contribute and maintain the code -- something you haven't got] I dont really see any contributions into Jboss kernel from parties other than Jboss employees . For that matter, I hardly ever see a problem being resolved on your mailing lists or forums by the user community on its own-- we all hope and pray for one of your guys to come around to resolving it. Most users, including me, are clueless about Jboss internals. There is a complete lack of third party documentation or books and your guys can't write(sorry Fleury etc.).

    Your software is no better that any other commercial software.You have lasted this long not because of Jboss community -- we are all too cheap for that -- but because your consulting arm generates business for you.

    I use Jboss because my customers are too cheap to pay for a license for even Pramati, Resin or JRUN. And I am too lazy to work with OpenEJB or learn JOnAS -- which is getting better everyday. There -- I said it.

    (If anybody from Paramati , Rsein or JRun is listening -- here is how you beat Jboss: collect royalties instead of upfront license fee)
  12. How is JBoss different?[ Go to top ]

    "I dont really see any contributions into Jboss kernel from parties other than Jboss employees."

    Is this not true for other OSS projects, as well? From the source I have looked at in some Java OSS, most of the "core" code is written by only a handful of people - usually the person who contributed the initial code plus a few others. As you move further away from the core code and into other modules, you see more and more contributors.

    I am not familiar with some of the other non-Java OSS successes, such as Apache or Linux. Is the code from the Linux kernel made from contributions of many developers, or just a handful of those tied closely to the project?

    As for saying JBoss is hardly open source... The source is distributed with the application, and it comes with the LGPL. How is this not OSS? Just because there is a close group of developers that drive the application's development doens't mean is isn't open source. Doesn't Torvalds shepherd the Linux kernel development as a "benevolent dictator". Is this not OSS, too?

    Ryan
  13. Jboss is barely open source[ Go to top ]

    Ryan

    Let me clarify: Jboss is open source as you pointed out but barely so. The key difference is that the successful Open source projects succeed in getting a wide cross-section of developers who belong to several different organizations and they also build up a wide body of knowldege that is available in the user community. In this case , all developers are Jboss LLC employees and the user community is pretty much clueless about the source code and archtechture, yours truly included.If you visit their forums or mailing list archives, you can see just how clueless we are.

    There is no third party documentation at all for Jboss while there are a gazillion books for Linux, Apache, Struts or Tomcat. Also, do you see organizations other than Jboss LLC. competing to sell support and services for JBoss App Server? Thats the litmus test for OSS and they failed it.

    So even though Jboss is open source, it is not in the same league as Linux or Apache/Jakarta projects.
  14. Re: How is JBoss different?[ Go to top ]

    I think that while OSS code bases can potentially be of higher quality vis a vis commercial software, the spending that large software companies can bring to bear on the research front is difficult to replicate in an OSS scenario (That's not to say that it's not possible - take a look at IBM)

    And when a company pours in a lot of money into research that ultimately materializes as software, I say it's upto that company to decide whether they want to share that code with anybody or not. And if they don't want to, well that's their prerogative. And rightfully so.

    On the other hand, IBM, for instance, realized that being a leader in Linux-based solutions space has distinct business advantages, and that Linux could bring great value to their customers. That's difficult to pull off though.

    In any case, there's space for several paradigms to harmoniously co-exist out there.

    Sandeep.
  15. Jboss is hardly open source[ Go to top ]

    Right on, Damien!
    I don't see any advantage in being able to debug let's say Tomcat. On the contrary, it is ridiculous: how many people would work with and deploy on there own version of Solaris or AIX or Linux for that matter? A large % of effort in a big project is spent on standardization. All of the sudden this is not true anymore?
    Might be an MS joke, but "Use the source, Luke" is very true.
  16. JBoss is OPEN SOURCE[ Go to top ]

    Posted By: DODO DODO on January 18, 2003 @ 07:16 PM

    > Right on, Damien!
    >I don't see any advantage in being able to debug let's say Tomcat. On >the contrary, it is ridiculous: how many people would work with and >deploy on there own version of Solaris or AIX or Linux for that matter? >A large % of effort in a big project is spent on standardization. All of >the sudden this is not true anymore?
    >Might be an MS joke, but "Use the source, Luke" is very true.
    >You don't see any advantage in being able to step into the source code >to see what is causing a problem eh?

    So you would rather send an email off to tech support or spend hours on the phone to resolve an issue you could have looked into yourself? You call yourself a developer?

    Standardization is still important... what does JBOSS being open source have to do with standardization? You standardized on J2EE when you decided to use JBOSS.

    Having the source code at your side in case you need it is just one of the things that make JBOSS and other OPEN SOURCE projects so great.

    The license alone makes JBOSS as an open source project. You are free to do with it what you will without penalty. Developers participate in it's evolution and there are now books available. Check out the JBOSS 3.0 Developers and Administrators Handbook from Wrox Press. It's a pretty good book.

    Damiens definition of "Open Source" simply does not hold water.
  17. JBoss.com?[ Go to top ]

    "Damiens definition of "Open Source" simply does not hold water. "

    It sure does, because JBoss does not operate like any other OS project - they actively attempt to make money from documentation/consulting services.

    It is no different from other commercial offerings - you can save money on licenses, but you will spend it on 'consultants'.

    I wonder why jboss uses .org domain - they are clearly trying to become a commercial company, so why not use jboss.com?
  18. nn nn,

    I actually see a very big difference between JBoss and commercial offerings. First of all, the fact that the JBoss group provides commercial support does not make the JBoss software less open source. Second, the availability of commercial support on JBoss makes it even more interesting. Third, the free support I got from the JBoss forums was better in quality and in response-time then any other commercial product I've used.

    Distributing development efforts under LGPL deserves respect because the developer risks that a commercial competitor takes advantage of his efforts.

    The only advantage that commercial products can have over OSS is the completeness of documentation and features. But in the case of JBoss I do not believe many projects exist for which the documentation and/or features are not sufficient. I think that one of the big reasons that JBoss is under-appreciated are the managers (without technical knowledge) whom believe their ass is better covered if they spend big bucks on big names. (excuse my vocabulary). Ben, correct me if I'm wrong.

    Of course, this is only my humble opinion / experience.
  19. Hi,

    I really like the OSS software and we - banking - are using a lot of them but I wonder if that model will work in future, too ?

    How are those OSS projects financed ?

    Will those OSS projects have the same energy they had the last years ?

    Now that everybody has to earn money again what's the direction OSS goes ?

    If I - today - rely on Tomcat, jBoss, Struts and even exchange jBuilder for an OSS plattform, what seems to be possible at the moment, what will be in two or three years, when our development unit heavily relies on all this stuff ?

    Any comments appreciated.

    Oliver.Lauer@sk-koeln.de
    Stadtsparkasse Cologne
    Germany
  20. Hi there,

    "what will be in two or three years?"

    these are very valid questions, but they are unfortunately very hard to answer. From past experience I'd say that you simply cannot rely on tools and technologies to stay around, except for some really big ones like C++, SQL, Linux.

    So, I'd guard against a changing world, using simple-sounding architecture ideas like:
    - modularized architecture. If one module fails or the
      underlying technology is discontinued, only the module
      has to be replaced
    - use open standards. If a vendor or OSS-supplier fails,
      there will be another vendor that implements the
      standard
    - use agile development processes. If "constant change" is
      the way the world works, make sure the developers
      know how to cope with it.
    - accept heterogeneity.
    - disbelieve the hype. Most "grand unifying architecture
      initiatives" are
      a) a marketing ploy to lock you into a particular
         system
      b) 5 years from now yet another legacy system


    Cheers,
        Henrik
        TNGtech
  21. JBoss != commercial[ Go to top ]

    "It is no different from other commercial offerings."

    I will give you one difference - the software is FREE. Commercial software companies sell you software AND try to sell you consulting services on top of it. But if you DO go with a commercial server, you do get the docs free. Yee hah, what a cost savings that is!

    But what about this scenario. You are starting a new application with some developers what already know JBoss well. Then there is NO cost. If you go the commercial route and you don't need consultants, you still have to buy the software.

    Plus, what the hell is wrong with trying to make money of this software? People have been making money on OSS for a long time. Unless you want to tell me that there are no people out there earning a living as Linux admins. All the JBoss group is saying is, "Here is some kick ass software for free. Do with it what you like. If you want some piece of mind, hire some experts from our group to help you get started. If you want some more piece of mind, we offer support contracts, too. But hey, if you don't want any of that, you can still get our kick ass software for free."

    Oh, and www.jboss.com is taken.

    And one more thing related to another post asking what will happen if you rely on OSS that may not be around in two or three years. Well, I would say that there is a better chance of JBoss and/or Tomcat lasting than some of these commercial app servers. And if JBoss development does die, you still have the source code. Can't say the same with BEA (no knock on WebLogic intended).

    Ryan
  22. !(JBoss != commercial)[ Go to top ]

    Ryan

    Jboss is a decent software and your source code is indeed available freely. But your community sucks and so does your documentation. There is absolutely no third party support or documentation available. I actively use your forums and mailing lists -- and I can vouch for the fact that your user community (and that includes me)is pretty frigging clueless. Your source code may be available but I dont see any evidence of a user fixing another user's problem and posting a solution. Your source code CD is like that T-shirt giveaways at conventions : nice to have but doesnt help me much.

    There is nothing wrong with you making money from Jboss and I am glad Fortune 500 guys are using your software. But all your users are not Fortune 500 and we cant find cheaper support anywhere. And your documentation sucks and I cant seem to find any decent third party documentation.

    Hear me again : no third party support or documentation for Jboss sounds as though your intention is to monopolize the support business.
  23. JBoss[ Go to top ]

    FWIW, I am not affiliated with JBoss. It's not "my community" or "my" documentation. I'm just an outside observer.

    BTW - have you looked at the JBoss source code? It is quite modular and a few hours of time will let you see how it fits together. Understanding JMX is a big help. (Actaully, reading the JBoss docs and the source code is one of the best ways out there to learn JMX). And since it is modular, you can look at each piece seperately to see what is going on.

    Ryan
  24. Dear Clueless[ Go to top ]

    Contrary to your clueless assertion that there is no third party JBoss support, what do you call "JBoss 3.0 Deployment and Administration" by Wrox Press.

    Looks to me like your real problem is your admitted cluelessness. And you get mad because nobody is coming along to spoon feed you the answers.

    My only advice to you is to wise up and get a brain so you can figure the software out, or get a few bucks together and take the JBoss training.

    I hate to break the news to you, but there is no free lunch, not even in open source. Either pay the effort to figure it out yourself, or pay $$ to someone to help you. Either way, there is no free lunch. And to expect otherwise is being really, really clueless.
  25. Wrox book[ Go to top ]

    <greg>
    Contrary to your clueless assertion that there is no third party JBoss support, what do you call "JBoss 3.0 Deployment and Administration" by Wrox Press.
    </greg>
    Good point. The author of that book wasn't a JBoss employee or even a contributor. However, it makes it all the stranger that Bill Burke of JBoss is trying to tell people not to buy that book in another thread on this site. I think selling consultancy around JBoss is absolutely fine; relying on selling documentation has the potential to cause frustration and reduce takeup of JBoss.

    I agree with Ryan: I find the JBoss code well written and easy to understand. As far as I'm concerned it's a true open source project and I think it's fantastic competition for BEA et al.

    Rod
  26. Jboss is barely open source[ Go to top ]

    Look I have been using Linux + KDE + Mozilla + NetBeans without attending any pricy training or looking at the kernel code. Why are there no decent third party books or training /support organizxation like we have for Linux/ Tomcat/Netbeans/Apache/SendMail etc. ?

    If there have been millions of downloads of Jboss , then why is it that O'REILLY or other publishers are not publishing titles on Jboss ? (I own WROX book and its a POS. That book is remarkable for its complete lack of any new insight.)

    I am not asking for a free lunch but only something more affordable and maybe a few choices. Seems to me that the Jboss organization doesn't like competition.



    -dd
  27. JBoss.com?[ Go to top ]

    FYI, I am guessing you are not that savvy, jboss.com has been registered sinse 1996 well before JBoss and J2EE ever stepped foot into the market.
  28. JBoss and other supported software ...[ Go to top ]

    <Q>
    It is no different from other commercial offerings - you can save money on licenses, but you will spend it on 'consultants'.
    </Q>

    Sounds like a big difference to me. Most software requires help. Why not spend money on qualified people as opposed to licenses and then only being able to afford entry level types.
  29. Invest in your OWN Intellectual Property[ Go to top ]

    The main message we send to our customer base is one of investing in their own people, not empty software liceses that still require training. If we collectively empower big business with the source and the knowlodge, they maintain control of their IP. Gone are the days of holding code in escrow, worrying about software companies going out of business, or an EOL on a software product. With OSS and JBoss specifically you can have the code from the beginning. Not all companies want to train their people to be experts in the product, which is why they pay JBoss Group for support. Not all companies that use JBoss pay for support.

    In general, most companies would be better off if instead of investing $$$$ in software licenses, took the same money or less in most cases and made their employees smarter. A few years back this would not have been as effective, as the job jumping craze of the dot com generation, caused many companies not invest in their people, as they might walk out the door tomorrow with 5 offers for 50% more than they were making. These days, job offers are not as plentiful for developers and admins, and thus, by investing in people and making them an important part of your infrastructure, you maintain control.

    What ever happend to people being the most valuable assest a company has? Is this no longer important?
  30. JBoss is OPEN SOURCE[ Go to top ]

    Standardization is still important... what does JBOSS >being open source have to do with standardization? You >standardized on J2EE when you decided to use JBOSS.


    So in your opinion a statement like: "We use SQL, unix and RISC machines" is standardization. Excellent!
    You completely missed the point: when you debug and fix something(e.g. your beloved Jboss) you branch out of the mainstream. The fix exists only on your machine. So you have two possibilities: spread your offspring to the client or submit it to the open source comunity. Both cases involve testing, regression testing and some bug fix admission process(I hope, for the benefit of your company). This means time. So the advantage of having the code means nothing, except that you spent time on other people's code(your project manager will give you a bonus for this... of course). Ever heard of the black box concept?

    >You call yourself a developer?

    I'm trying to ignore this question. It seems that your definition of a developer involves to much jolt cola(at least).
  31. Source code does help.[ Go to top ]

    So not having the source code is better? If you have a problem with a closed-source software, you

    A. Have no idea WHY the bug is happening.
    B. You have to wait for them to find out WHY the bug is happening
    C. You have to wait for them to give you a patch.
    D. Until then, you are screwed.

    Explain to me how this is better again, because I missed how having less control is better.

    FYI, we modified some OSS last week because it didn't exactly suit our needs. Having the source code is a benefit.

    Ryan
  32. Souce Code is no magical Panacea[ Go to top ]

    Even though access to source code helps, it is often useless in time-critical situations. How many times have you checked your Linux Kernel code to see why your USB driver is nor working ?

    Having access to source code is important and reassuring , but having a knowldegeable supportive user community with a ton of documenation is just as , if now more, important.

    There are Users and developers: typically developers extend the OSS to embed in their commercial wares aka Linux + IBM or Tomcat + IBM/Sun etc.

    Then there are users like me -- I use the OSS as an end in itself: I am not extending Tomcat or Linux -- just using it for my specific purpose. And while most user can do without a lot of support , we do need good quality documentation.

    -dd
  33. Souce Code is no magical Panacea[ Go to top ]

    Then there are users like me -- I use the OSS as an

    > end in itself: I am not extending Tomcat or
    > Linux -- just using it for my specific purpose.
    > And while most user can do without a lot of
    > support , we do need good quality documentation.

    Right. Life is too short for digging into ugly source code. It's just a matter of cost. If I need 10 days to set up JBoss, because I have to dig into the source, fight flame wars in JBoss forums, it cost me 10 * <daily rate> plus the rate of my psychatrist to recover me from te flames. In that case it would certainly be a better and more cost effective way to purchase e.g. Resin or JRun or use JOnAS. At the end of the day it should just work, reliable.

    If I need new software, the first I do is to look for *commercial* products. There are a lot of companies out there which offer nice commercial products with excellent support (if requires at all).

    One example, a couple of weeks ago I was looking for a new Linux mail server for our company to replace our old NT based stuff and to retire the NT server. Linux works, no doubt. So I was looking at sendmail, postfix, etc. We needed web mail and mailing list as well. So I looked to various open source web mail clients and mailing list managers. It is horrible to set everything up to work together. Some packages are only configurable through command line etc pp. That the source code is available is completely out of interest for me. I would never change the source to work around obstacles. Thus, at the end I found I'm looking for. A nice, feature rich mail server with everything I need with a top notch admin tool. Paid a few hundred dollars by credit card, downloaded, installed within 5 minutes and works without problems. It's Kerio Mailserver.

    This kind of software and software companies will have success now and in future. That's a business model which works, because it's honest. Open source or not, who cares. I guess most of the 2 million downloads of JBoss came from themself or from India and China. But certainly not from Fortune xxx companies. That's all blah blah. JBoss seems for me like a religion with a megalomaniac guru on the top. Want more, want rw on cvs? Book your next training! Become a Rock Star! LOL.

    -- Andreas
  34. Find a different job[ Go to top ]

    Andreas,

    If you need 10 days to set up JBoss,
    find another job.

    gr,

    Jan
  35. Find a different job[ Go to top ]

    If you need 10 days to set up JBoss,

    > find another job.

    You must be a Rock Star!
  36. RW access[ Go to top ]

    "Want more, want rw on cvs? Book your next training! Become a Rock Star! LOL"

    Andreas - have you tried to be a committer on this project? Do you even no how this process works? It probably works just like any other OSS project:

    You start of by getting familiar with the source code. Then you maybe submit a patch for a bug/enhancement you discovered. Then you do a little more of this. Then perhaps you start writing some new code from scratch and submitting it. Then you trade emails with some of the main project people about improving the project. Once you build enough trust and prove you are your value to the project, you WILL GET rw access the CVS tree. Why should it be any other way?

    Or have you already done this and been jilted. I doubt it.

    BTW - how many people have rw access to Tomcat? Linux? MySql? Apache Web Server? Ant?

    Ryan
  37. RW access[ Go to top ]

    Andreas - have you tried to be a committer

    > on this project?

    God beware! Fixing their JMS XA/ASF stuff to make it spec compliant and to integrate SwiftMQ was quite enough for me.

    I have done the very same work with the JOnAS guys. That was real cooperation. That was a pure joy.
  38. "Even though access to source code helps, it is often useless in time-critical situations."

    So not having the source code is better? When would not having the source code be more advantageous. I still don't understand your point. Besides, I have found that commercial software support can be equally useless in time-critical situations.

    Oh, and Andreas - I would not call most OSS's source code "ugly". Most of it (including JBoss) is quite elegant and a good lesson in the use of abstractions to make modular software.

    Ryan
  39. Ugly OOS code?[ Go to top ]

    Would you like some UGLY code from closed sourced vendors that shows how to use their products? I've got some.
  40. Ugly code[ Go to top ]

    Mark,

    This sounds like a very generous offer, but I will have to pass. I already have my share :-)

    Ryan
  41. Source code does help.[ Go to top ]

    From my experience in large corporations saving money has been accomplished through technology standardization, great negoiated contracts, and reduction of resources. Standardization takes place to get rid of the crap brought in by every rogue group in a company and determing the root cause is usually a developer or contractor trying to solve their own little problem without seeking professional help. Large companies negotiate reasonable software costs for big purchases, this usually covers a broad number of applications not just a few. People become the biggest asset in I/T and open source solutions just add to the cost of those who don't standardize. Sure small companies benefit with OS where a developer can wear all the hats in building and supporting an application but large companies break down these barriers by adding people. So in large companies OS is not free at all and anybody who has to integrate these solutions will find out managing these products requires yet more development.

    The comments so far bring up some great points and one that stands out is what types of systems are being created and for how many people? Sure if I had to support an internal application for 200 people at 10/tps then OS would probably fit the bill. Much different than say 100k concurrent users where a commercial product has most likely been proven, that's why I'm bringing out the checkbook, reduced RISK. Everybody can make promises but few meet the bill when scalability becomes an issue and that's why IBM, BEA, Oracle and others will continue to have people spending big $$$. The last thing a CxO wants to hear is "20 year old johnny is running though the source right now but were only losing $100k/hr until the application is patched".

    Totally agree on most solutions are over-engineered with the wrong software for the actual need. I've come to the conclusion that most people don't realize the power of a product until your properly trained, this should be a pre-requisite for evaluation instead of RFP's.

    The comment about having the source code available to fix problems is silly, how many people actually have the time and people with a mission critical environment to debug a complex product like jboss? Every contract I've seen within large companies have it spelled out if the vendor goes out of business you get the source code. If not then hire some smarter lawyers.

    It's interesting to see that most applications end up finding their database being the major bottleneck but I don't see a solid OS competitor that matches up to Oracle or DB2. Talk about high costs, it makes app servers look like buying candy. I would like to see mySQL or postgre handling the high loads and flexibility of deployment that Oracle does, I'll switch any immediately. I like OS solutions if the internet application is non-revenue based but I would be leary of high volume revenue based sites on OS.

    As for Jboss down the road I would potentially be worried like most OS solutions. It's just a matter of economics concerning supply, demand, and the good ole greenback. So let's say jboss begins to standardize in large companies because that's where the real money is, how can jboss end up supporting the tough demands of these companies without a real revenue stream? Unless they figure professional support is adequate (lots of people) then I would suspect jboss will not be free down the road. Of course the more devious path would be BEA or IBM to just buy the jboss guys out and end their future, gee what's a few million dollars from having their sales people constantly convincing those more informed technology decision makers from choosing jboss. Of course BEA and IBM could just lower their costs if demand starts to slow which will what probably happens. Oracle is the one to watch, if your already paying huge dollars for their db, why not just throw in all the other pieces for free? Now I have free IDE's, O/R tool, portal, etc, yet they stay on course with their primary product selling database software.

    OS solutions may be free but nothing is ever free down the road.
  42. "usually a developer or contractor trying to solve their own little problem without seeking professional help"

    Very funny.

    Without all this thousands of, shall we call them "power-user" solutions?, the big companies would grid to an halt immediately.

    Contrary to the big Java app server projects (done by so called "professionals") which fails in 80% of the cases, according to Gartner.

    Regards
    Rolf Tollerud
  43. ... according to the CHAOS report 1994 and there was no Java then.

    Figures have changed a bit (http://www.softwaremag.com/archive/2001feb/images/ProjectSuccess.jpeg) but the rate of failed or challenged projects is still around 70%. Microprojects are the trend for 2003 (according to Standish Group) and are often more successful than big projects, but they bring their own set of problems, notably in terms of management and reuse.

    By the way, there are many successful J2EE-based microprojects around. The high failure rate of huge projects has nothing to do with J2EE application servers, it is more a question of poor communication, political issues (think of those many buyouts between banks and, insurance companies who were former competitors and who are often provoked to merge their heterogeneous systems), lack of a good understanding of all businesses involved and irrealistic targets (think of Microsoft desesperatly trying to get Hotmail on NT servers for years and finally giving up because the architecture would not scale). Standardisation (huge failure projects are often large standardisation or federation attempts), in order to be successful, requires a strong and stringent process. Microprojects on the other hand are easy to manage and the business scope is localised.

    Microprojects will eventually become collections of business processes available to higher-level management systems such as BPM engines. That's probably a more secure approach than getting directly into huge projects.

    But J2EE application servers have nothing to do with failure rate.

                    Yann
  44. Yann,

    Agreed. The main part of large IT-projects fail, whatever platform.

    But,

    The big Java app servers are build for and used mainly in large projects. The people who is behind products like Websphere/Weblogic is the same (kind of) people that advocates the large projects.

    Regards
    Rolf Tollerud
  45. Java app server success stories[ Go to top ]

    Yann,

    BTW, Hotmail is running on windows 2000 now..

    And for every large Java app server success story you can seriously document, I will give you 10 MS projects of the same size or larger.


    Regards
    Rolf Tollerud
  46. Java app server success stories[ Go to top ]

    "BTW, Hotmail is running on windows 2000 now.."

    Indeed. Just read a Microsoft paper that confirms it. It only took the company with the highest IQ/m² company 3.5 years to put it all together. And I was wrong, that was not initially a Solaris system but a FreeBSD. That is not exactly a success story.

    "And for every large Java app server success story you can seriously document, I will give you 10 MS projects of the same size or larger."

    I'm afraid there is no way you could back this assertion with facts. For one, neither you nor I know all large projects in the world to generate statistics. Second, there is no way to know whether those projects were failures or successes. Third, there is no way to know the real reason why projects failed. As I said, most of the time, technology has nothing to do with it. And so much for statistics anyway...

    I'm convinced any project has the potential to be a success regardless of the technology. The rest is just marketing, hype, FUD and religion. One of my duties is to read through that to agnostically use technology where it best fits. For instance, I'm not an entity bean fan at all because I've seen it kill performance in numerousous occasions so I tend not to listen to Sun's or BEA's recommendations. But you can be sure that XML-based value objects for inter-tier communication will kill performance on J2EE, .NET, or any other business platform.

    Perhaps in your experience, you have seen many J2EE projects fail. I have seen a few failures too, all of them were for political or unrealistic targets reasons. But I have objectively mostly seen successes in J2EE, sometimes with slipped deadlines, sometimes with overkill engineering, sometimes not. The world is not a perfect place but the solutions worked as expected. All I know is that I feel pretty confident that Java/J2EE covers most of the current IT needs and I'm an overall happy Java/J2EE developer/architect. After all, the only objective is to keep customers happy and meet the specified requirements. I can do that with Java. I could probably do it with C++, VB, C#, ColdFusion or PHP but I'm in Java and many customers trust Java because they are confident it is one of the best technologies around at the moment with a large developers community. Other customers are Microsoft shops, others only want OSS, other want their good old C. It's all fine, there's room for everyone. The rest is just politics.

                    Yann
  47. Microsoft IQ/m²[ Go to top ]

    Yann,

    "It only took the company with the highest IQ/m² company.."

    It seems that this is still the case according to this survey,
    http://home.techies.com/Common/Content/2002/12/15mc_idealemployers.html

    Notice that the reason that people want to work for the next highest, IBM, is "stability", but the the reason they want to work for MS is "Interesting, challenging work". You can draw your own conclusions from that.

    "Most Desired IT Employers of 2002"

    Microsoft 4.8%
    IBM 4.7%
    U.S. govt. [military or civilian] 3.6%
    Cisco Systems 3.5%
    Government [local or state] 2.2%
    AOL-Time Warner 2%
    Sun Microsystems 1.9%
    Dell Computer 1.8%
    Walt Disney 1.8%
    Lockheed Martin Corp. 1.6%
    Apple Computer, Inc. 1.3%
    AT&T Corp. 1.2%
    Hewlett-Packard 1.2%
    Charities, religious orgs., nonprofits 1.1%
    The Coca Cola Company 1%
    Intel 1%
    3Com 1%
    NASA 0.8%
    Verizon Communications 0.8%
    General Electric Co. 0.8%


    For IBM 19 percent of techies say the company's stability is the biggest pull. Although reasons for choosing IBM align closely with those offered for all firms totaled, this is not the case for Microsoft and Cisco. "Interesting, challenging work" is the biggest attraction to Microsoft [19 percent] and Cisco [25 percent], followed closely by "cutting-edge technology" [19 percent and 24 percent, respectively].

    Regards
    Rolf Tollerud
  48. Microsoft IQ/m²[ Go to top ]

    Rolf,

    It's worth working for any company that can whet your appetite and help you learn new things. Microsoft is probably a great company as far as competition is concerned: it has a strong identity and being the leader in many areas and working to keeping this position encourages team spirit. The J2EE/.NET rivalry can also be perceived as very motivating for both sides. I'm pretty sure people working for Sun, Oracle or BEA would say the same thing overall. IBM is different because it has existed for such a long time that it has become an institution, and inspires security. There's no conclusion to draw from that.

    One thing is certain, it is sure as exiting to work at the cutting edge of Java than it is to be at the cutting edge of .NET, C++, OSS, jazz, modern art, sport, politics, carpentry or any activity you do with a reasonable bit of passion. I don't think a Redmond engineer is more motivated that her Palo Alto counterpart.

    As I said in other posts, the rest is just political manipulation and biggots of any camp are merely likes of Jeovah Witnesses trying to convert people to their own camp for your own good of course. Just football team spirit. A bit of hindsight does no harm... I know what I can and can't do with Java/J2EE, and that's enough to satisfy most of the customers I work with.

                    Yann
  49. Java app server success stories[ Go to top ]

    "BTW, Hotmail is running on windows 2000 now.."

    Last I heard after the conversion from FreeBSD -> W2K was it went from 10,000 servers to 20,000 servers. Oh and the backend still requires 200 fully loaded Sun e6500's. But when you have 40b in the bank I suppose they don't care, it's all about marketing. I've heard that datacenter is absolutely huge..


    Ryan..
    "Even though access to source code helps, it is often useless in time-critical situations."

    My point was that time needed to invest in people to truely understand the source code is difficult to justify. Most companies will resist this pattern, it's all about reducing risk. Much easier from a legal standpoint to go after a company than a community if something fails miserably and it's costs your company. Also it's not easy at times to hire highly compentent people, I'm sure not every company has a "Rod Johnson" or similar available to solve issues everyday.
  50. Java app server success stories[ Go to top ]

    "Much easier from a legal standpoint to go after a company than a community if something fails miserably and it's costs your company."

    I hope companies aren't basing their software choices on who will be a better scapegoat when things go wrong. I understand risk mitigation is an important factor in software selection, but the question should be "which works better", not "whose ass can I chew when things don't go right". Also, there is some risk involved with closed-source as well. When happens when you app server vendor goes belly up, or when they stop supporting your product. With OSS, at least you own the code.

    Also, you don't really need to know the souce code to use an OSS product. I have used several and have never needed to crack open the source code.

    "Also it's not easy at times to hire highly compentent people."

    Agreed. But incompentent people can screw up a commercial solution just as easy. And sometimes the people supporting these commercial product are idiots, too. Which goes right back to an OSS mantra - spend money on people, not software. If you hire talented developers, you will get quality solutions, OSS or no OSS. If you hire idiots, you will get crap, no matter how many "wizards" your commercial software offers.

    Ryan
  51. Hope against all Hope[ Go to top ]

    <I hope companies aren't basing their software choices on who will be a better scapegoat when things go wrong.>

    I hope lots of things like this.

    Like: I hope that companies/orgs keep the employees/contractors/whatever based on qualifications, skills and abilities, not who has been there the longest.

    Like: I hope companies/orgs really do want good, quality software.

    Unfortunately my hopes are being constantly dashed. :(
  52. The site seems to indicate MSFT still uses solaris for backend servers ..
    http://www.theinquirer.net/?article=6900
  53. Java app server success stories[ Go to top ]

    "BTW, Hotmail is running on windows 2000 now.."

    Last I heard after the conversion from FreeBSD -> W2K was it went from 10,000 servers to 20,000 servers. Oh and the backend still requires 200 fully loaded Sun e6500's. But when you have 40b in the bank I suppose they don't care, it's all about marketing. I've heard that datacenter is absolutely huge..


    Ryan..
    "Even though access to source code helps, it is often useless in time-critical situations."

    My point was that time needed to invest in people to truely understand the source code is difficult to justify. Most companies will resist this pattern, it's all about reducing risk. Much easier from a legal standpoint to go after a company than a community if something fails miserably and it's costs your company. Also it's not easy at times to hire highly compentent people, I'm sure not every company has a "Rod Johnson" or similar available to solve issues everyday.
  54. Java app server success stories[ Go to top ]

    Joe,

    "Last I heard after the conversion from FreeBSD -> W2K was it went from 10,000 servers to 20,000 servers. Oh and the backend still requires 200 fully loaded Sun e6500's."

    Where do you get this rumors and old wife tales from? I can asure you that Hotmail is running only on Microsoft boxes.

    Moreover, when Hotmail was bought and running on FreeBSD, Java was not envolved. In fact, there is not a *single* high load mission critical J2EE server solution in the whole world. I would be glad to be contradicted, just name me one.

    "EJB is FUBAR" (f***ed up beyond all repair)

    Regards
    Rolf Tollerud
  55. Java app server success stories[ Go to top ]

    Rolf,

    "I can assure you that Hotmail is running only on Microsoft boxes."

    How can you be so sure ? Did you visit their data center ?

    "In fact, there is not a *single* high load mission critical J2EE server solution in the whole world. I would be glad to be contradicted, just name me one."

    Just one then: http://www.fxmicropay.com/?page=fxmp_media. Royal Bank of Scotland. Unless you say 3.000 transactions/second is not a high load mission-critical...

    However, I don't believe J2EE is best suited for highly-critical transactional environments such as money clearing for instance. There you'd need something like Tuxedo. But, to be honest, such systems only represent a very small percentage of the applications and J2EE application servers do fine in most circumstances.

    "EJB is FUBAR (f***ed up beyond all repair)"

    Just another personal provocative empty statement.

                    Yann
  56. Java app server success stories[ Go to top ]

    Yann,

    I do believe you but when I go to the site I see only asp pages, and Netcraft is reporting:

    The site www.fxmicropay.com is running Microsoft-IIS/5.0 on Windows 2000

    Please enlighten me.

    Anyhow so is the National Bank of Costa Rica converting their entire systems of 2.5 mill rows of code to .NET/C#. I have to get me an account in both banks then so I can compare the user experience!

    "EJB is FUBAR (f***ed up beyond all repair)" "Just another personal provocative empty statement."

    It is not I who said it, but Vic Cekvenich. In another thread in this forum,
    http://www2.theserverside.com/home/thread.jsp?thread_id=17497&article_count=55#71439.

    Vic is a 100% Java guy who wrote the book:
    Struts Fast Track: J2EE/JSP Framework: Practical Application with Database Access and Struts Extension.

    President for Basebeans "saving crashed Java projects" Engineering.
    http://www.basebeans.com.


    Regards
    Rolf Tollerud
  57. Java app server success stories[ Go to top ]

    Rolf,

    "The site www.fxmicropay.com is running Microsoft-IIS/5.0 on Windows 2000. Please enlighten me"

    LOL! I was actually referring to the contents of the page. :)

    And I did not say it was you for the empty provocative statement. I just said that such a statement brings nothing to the table. It's not a constructive criticism in other words. It's not worth being emphasized, especially taken out of the context it was written in. It sounds like a rant. And whether it comes from a 100% Java guy who wrote a book on Struts, or from a 100% .NET, it is only one opinion, with which I disagree: parts of EJBs are really nice. I agree it can be complex, but reality is a complex thing and part of the difficulty in critical systems resides in managing the many unseen alternative flows in your use cases and providing documented and proved answers to the "what happens if it crashes at that moment ?" question. Technologies are equivalent in front of these problems. And you have to know when not to use them.

    Rolf, you seem to be pretty active trying to scare people away from Java to .NET. Why that ?

                    Yann
  58. Java app server success stories[ Go to top ]

    Yann,

    "Rolf, you seem to be pretty active trying to scare people away from Java to .NET. Why that?"

    I do not care a bit about the Java programmers, only to stop companies from using J2EE which I think is badly flawed.

    The main reason I do it though, is for entertainment! It is always great fun to stop a needle in people with inflated though about themselves and their work.

    I don't count you in among these people though..

    Regards
    Rolf Tollerud
  59. Java app server success stories[ Go to top ]

    "I do not care a bit about the Java programmers, only to stop companies from using J2EE which I think is badly flawed."

    By J2EE do you mean Java or J2EE or EJBs?

    "The main reason I do it though, is for entertainment! It is always great fun to stop a needle in people with inflated though about themselves and their work."

    So you are like the court jester?

    "IBM (and a bunch of other companies) tries to destroy the market for Windows with OSS/Linux. Microsoft likewise destroys the market for Java by giving .NET to OSS/Mono.
    Illusions anyone? "

    Hmmm.

    Before discussing illusions let's put 'Microsoft destroys market for Netscape by giving IE away' before IBM/Linux.

    As for illusions - Microsoft has not given .Net to OSS/Mono. If you define .Net as the CLI and C# - ok (but not specifically to OSS/Mono). But not Winforms or ADO.Net or ... in other words the bulk of .Net. On top of that not an IDE. If I remember correctly, that doesn't bother you. But it will 99% of those who will use .Net.
  60. Java app server success stories[ Go to top ]

    Rolf,

    "I do not care a bit about the Java programmers, only to stop companies from using J2EE which I think is badly flawed.

    The main reason I do it though, is for entertainment! It is always great fun to stop a needle in people with inflated though about themselves and their work."

    I suppose you work as an architect and have thoroughly surveyed both architectures to the point where you can describe the pros and cons of each technology. If that is the case, I respect your opinion. And I know quite a few people doing the same thing in the other direction too because they believe .NET cannot hold its promises. It's all a matter of perception anyway... I for one like Java because:

    - I know it quite well;
    - I can use it effectively in basically any context where I see fit;
    - I can be pretty accurate in setting my deadlines and meeting them;
    - I can tell realistic objectives from unrealistic ones;
    - I can usually deliver and meet the business and non-business requirements.

    Programming languages are called languages and it's a great metaphor. Say Java is English and C# French, I can say the same things in both languages. Some of those things will take 3 words in English and 10 in French to express, others will do just the contrary. Comparing languages or J2EE/.NET is like saying English is better than French - and I know quite a few who will strongly support that :)). Some people would say that English is spoken in the whole world whereas French only in countries of the francophony. Some others would argue that French is harder to learn than English. Some others would say that English speakers don't want to learn any other language, or that French should be the international language or that Esperanto is the magic solution to all our problems. Yet the reality is that French and English are both best where they belong.

    J2EE is certainly flawed in certain areas just like basically any other consortium-based standard because trade-offs needed to be made. On the other hand, the standards once they're achieved get a great worldwide attention and support - and criticism. There is a perception that J2EE standards are driven by actual enterprise needs, and they are to a great extend (even though mistakes are unavoidly made). Also, the perception is that Microsoft standards are driven with monopolistic views, which scares many people.

    I for one tend to think that the choice of technology should be driven by the requirements, the environment and the expertise of the people involved. Should I have to architect a solution for a company that is a Microsoft shop, I would probably recommend .NET and be a happy developer too. Should the solution be for a clearing system, I would probably go for the Tuxedo suite. The J2EE/.NET rivalry is artificial but it gives a sense of belonging to many, from both camps... football team spirit.

    If it entertains you to trigger flame wars just like an Arsenal supporter going to a pub in Liverpool wearing the Arsenal official T-shirt, fine. All you do is creating tension, which is not very constructive.

                    Yann
  61. Java app server success stories[ Go to top ]

    Rolf,

    I usually stay away from your childish garbage but this one is too much!

    If you actually think that because Netcraft returns Microsoft-IIS/5.0 on Windows 2000, that they're only using that, then you really must not know as much about web development and systems as you claim.

    I've worked for at least two companies that used J2EE app-servers on NT/2000 with IIS as the web server.

    Take my company, for instance. it returns "Apache/1.3.27 (Unix) PHP/4.2.3 mod_ssl/2.8.12 OpenSSL/0.9.6e".

    But, I'm pretty sure (actually quite positive since I'm somewhat responsible for it) that we are running JRun 4.0sp1 on that server.

    Sounds like you've got some critical and fairly basic development and systems conventions to learn.
  62. Java app server success stories[ Go to top ]

    Rolf, if you are not happy with J2EE you are certainly welcome to use .NET and write your apps in VB.Net, C#, COBOL, APL, PL/I, or any of the other 30 languages it supports. Before you drifted off into the J2EE sucks tangent, everyone was discussing the use of OSS J2EE in the corporate development space which is where I live. We made this decision three years ago. .NET was vapor and even today Microsoft cannot tell you what it is (http://zdnet.com.com/2100-1104-948838.html)

    My choice was based on good quality code, running across multiple platforms and using open standards. What good is writing in a hodge podge of languages that run on only one proprietary platform (do you really count half-baked Mono)? You would never be able to sell that in a heterogeneous enterprise environment. Also, if you want to improve your J2EE skills, so your code doesn't suck, try reading up on EJBs as natural caches (http://www.jboss.org/blue.pdf). My experience is that J2EE doesn't suck if properly implemented.
  63. Rolf,
    maybe we can ask Floyd to create a seperate page on which to place your repetative arguments. It could save you so much time and you could instead go to the pub.

    + "J2EE is crap" "If I didnt say it Vic Cenkovic did on thread X..."
    + "90% of Java projects fail..."
    + "Show me one large scale site..."
    + "Windows is fantastic..."
    + "Big Appservers are a disaster..."

    I find it remarkable that despite the emphatic put-downs that all of these arguments have repeatedly received, you still persist with them. They are always the same. Its like groundhog day.

    And then, due to the deliberately provocative nature of your posts, half the thread is taken up with posts like mine here because people get pissed off with you.

    -Nick
  64. Nick,

    If there was nothing in my (and Vics and many others) arguments, and "emphatic put-downs that all of these arguments have repeatedly received" was true,

    you wouldn’t be so angry..

    "Java is a beautiful and elegant language that have been kidnapped by EJB Orchs"

    Regards
    Rolf Tollerud
  65. Its not that there is much validity in your arguments, its just pure repetition. Its not that there is much validity in your arguments, its just pure repetition. Its not that there is much validity in your arguments, its just pure repetition. Its not that there is much validity in your arguments, its just pure repetition. Its not that there is much validity in your arguments, its just pure repetition. Its not that there is much validity in your arguments, its just pure repetition.

    (annoying isnt it?)

    -Nick
  66. Java app server success stories[ Go to top ]

    Rolf,

    I never indicated that the Sun boxes were running any Java code, in fact the Sun hardware handles the database/transaction support I believe under FreeBSD. I guess Microsoft would need a boatload more boxes to emulate the Sun workload. I get my rumors directly from Microsoft and Sun thank you.

    “In fact, there is not a *single* high load mission critical J2EE server solution in the whole world.”

    I don’t have any specific companies off hand but Schwab’s environment looks pretty high volume, I’m sure BEA and IBM could name a few. Should I assume all these mission critical apps are running on Microsoft solutions and we should be all consumed by that mentality again?

    If you’re out here to push buttons about what this web site represents only to offer little then it would probably be more productive for you to sell you ideas on a street corner. I think most people here would rather use their time promoting something positive.
  67. Java app server success stories[ Go to top ]

    there is not a *single* high load mission critical J2EE

    >> server solution in the whole world. I would be glad to be
    >> contradicted, just name me one.

    Rolf,

    I can name dozens that I personally had project oversight responsibility when I was Chief Architect of the NE for Sun Professional Services. Virtually every top tier financial institution, virtually every major insurance company, virtually all the top tier on-line retailers, virtually every fortune 10 to fortune 1000 company in the NE, have high-load mission critical apps running on J2EE, both on their intranet and the WWW.

    So, I guess your right, I'm unable to refer to a *single* instance in the "whole world", not even in just one state as small as RI (CVS for example)!

    -ernie
  68. Ernie,

    Am I to risk another post? (it is dangerous).

    I have never (unlike some) said that I am objective. I am a 100% .NET/mono supporter.

    "Virtually every top tier financial institution, virtually every major insurance company, virtually all the top tier on-line retailers, virtually every fortune 10 to fortune 1000 company in the NE, have high-load mission critical apps running on J2EE".

    This is what has been official accepted for a while, even by the MS world (because of the more powerful computers). For a while ago I began to suspect that there were something rotten in the state of Denmark, I am convinced that the failure rates is catastrophic and performance lousy by the few projects that finish.

    One thing is sure, it is virtually impossible to find some big Java server success project.., but time will tell, "You cannot fool all the people all the time".

    But I have to stop; I am not accustomed to personal attacks.

    Regards
    Rolf Tollerud
  69. Evidence[ Go to top ]

    "This is what has been official accepted for a while, even by the MS world (because of the more powerful computers). For a while ago I began to suspect that there were something rotten in the state of Denmark, I am convinced that the failure rates is catastrophic and performance lousy by the few projects that finish."

    I think you are so attached to this idea that any efforts to convince you otherwise, any examples (from the people who built the systems, no less), and any satisfaction with J2EE from the users and the business community (that would be me) would just be discounted by you.

    There is no proof strong enough to satisfy you.
  70. One thing is sure, it is virtually impossible to find some

    >> big Java server success project.., but time will tell,
    >> "You cannot fool all the people all the time".

    I don't get it Rolf. In my previous post I made it very clear that I was personally involved in dozens of such projects. I named one company for posteriority sake, if you want I'll name all the projects that I've personally led or had oversight responsibility for, that were successes and are still in operation. However, I get the feeling that even if I brought you to the company data centers to observe the bits flying you would still be in denial.

    During my tenure with Sun I dealt directly with scores of CXOs in fortune 10s to 1000s to assist them with tactical issues as well as strategic planning for mission critical infrastructures. In the vast majority of cases (>90%), Microsoft based infrastructures either played a minor role or no role at all. It's true that Microsoft based apps generally dominated the desktop but rarely were trusted for mission critical uses.

    -ernie
  71. Ernie,
     
    Microsoft based infrastructures either played a minor role or no role at all. It's true that Microsoft based apps generally dominated the desktop but rarely were trusted for mission critical uses.

    Is that so? Here you have 200 success stories to start with.
    http://www.microsoft.com/resources/casestudies/search.asp?ProTaxID=1287&InPTaxID=All&SolTaxID=All&CompSize=All&NoMonths=All&submit1=Submit

    "You can not fool all the people all the time"

    Regards
    Rolf Tollerud
  72. Isn't it interesting that all the lead stories are about foriegn companies, or companies not in the fortune 1000, or Microsoft sponsored/hosted sites, or that they involve external websites or the prjects that are "in-progress", ie., no success yet.

    The only person getting fooled "All the time is you my friend"!!

    -ernie
  73. "Java app server success stories"[ Go to top ]

    |
    | But I have to stop; I am not accustomed to personal attacks.
    |

    If only you would stop.

    You will be quoting "success stories" like this before long : http://www.webservile.com/stories/downwell.htm

    "putting silly things in quotes doesnt make them any more significant"

    -Nick
  74. no illusions[ Go to top ]

    I've been at the site www.webservile.com, and they seem to have some problems, as these qutations show:

    "It is difficult not to write satire." Working in the software industry, and in particular in the much-hyped Web services space, we feel this way a lot"

    "We don't have any illusions of grandeur about our talent.."

    Do you think you can arrage a meeting? I humbly offer myself to help.
    (for a fee of course)

    The Java community on the other hand do not need any help. They manage very well themselves.

    "Virtually every top tier financial institution, virtually every major insurance company, virtually all the top tier on-line retailers, virtually every fortune 10 to fortune 1000 company in the NE, have high-load mission critical apps running on J2EE"

    What a perfect example of satire! Even Jonathan Swift would be jealous.

    Regards
    Rolf Tollerud
  75. Source code does help.[ Go to top ]

    Ryan,

    I agree 100% with your depiction of the fundamental problem we face with closed source software (ie if you have a problem you are stuck until the vendor fixes it). I guess thats why we are starting to see an emergence of Commercial Software that also ship src code as insurance (e.g. Jira).

    However, the insurance that the source gives you has to be tempered with the ultimate desire to completely avoid these problems in the first place. Ie, if the software is of good quality, you shouldnt need to look at the src OR phone support.

    The reality is that just as there is good and bad commercial software, there is good and bad OS software.
    Open-source software is not automatically a win. (but in many areas, its winning more often)

    -Nick
  76. Jboss is hardly open source[ Go to top ]

    Damien,

    Please only state facts that you know are facts. We are completely open source and the community at large has in fact developed JBoss. It was not until 2002 that there was ever an employee hired to code JBoss specifically. And at that timem there was already version 3.0.
  77. JBoss is used by a Fortune 100 company for mission critical apps. Because of the sensitive nature, I can't say what company or why, but JBoss has proven to be "good enough". When I say good enough I don't mean that in a derogatory term, I mean it in a practical one. Other servers provide a lot of bells and whistles, but we don't need that. We need an EJB server that you can easily download, execute run.bat, and hot deploy a jar file. JBoss gives us that.
  78. OSS vs. Commercial[ Go to top ]

    I am also a happy open source user, but I would like to find out how often do the developers check out the source code.
    I like the concept of the source code being open to verification by developers, but also find it conflicting with OOP principle of hiding the implementation. When I was training in OOP skills, I learnt that as an OOP professional, I must not concerned with only the interface, and not the implementation. I am developing a financial application at a major bank, and I use Apache, Ant, and other open source tools. But I just don't see a reason why I should be able to review the source code, that is not what I am paid for.
    Let an independent body have access to the source code,
    Why cannot we leave software testing to an independent body like the FDA for drugs, NTSB for transport safety. Why must developers from every nook and corner of the world be involved in this initiative. Let developers design the tools, and leave the testing to a body that does just that. Being cost effective is good, but when I realize that my bank is going to save millions by moving to Linux, and the bank would not contribute anything to the open source community, I find all this a terrible waste of talent.
    Open source may be helping a lot of companies to cut costs, but how many of those are giving back ?
  79. OSS vs. Commercial[ Go to top ]

    Murali,

    OOP principles are merely principles. And sometimes you have to break the principles for efficiency. I too want to be concerned with the interfaces only. Yet, it happens that things don't work as expected, in which case you either contact support (which is not always as lengthy as many people say), or you can chase the problem yourself. Or both, which is recommended if you face a blocking problem. OSS is great in that respect because you can very easily trace specific problems, either through debugging tools or through manual processing. This has the advantage of helping you understand the technology better. However, when support for commercial products is not very good, decompilers such as Jad do a *very* good job. As a developer, you're paid for developing and solving technical issues when you face them. The answer is always in the code. And if you find a bug and a workaround, that saves the OS team all the time you spent.

    Now, when a bank saves millions using Linux/Apache, I find this good, simply because those millions are then spent on other things such as new projects, or help avoiding axing jobs or bankcrupcy, or more likely company help shareholders buying boats or houses after quarterly results contributing to money redistribution. :) A true waste of talent would be that the hours spent writing OS are simply lost for nothing. Good OSS is good for everyone.

                    Yann
  80. OSS vs. Commercial[ Go to top ]

    <yann>
    When support for commercial products is not very good, decompilers such as Jad do a *very* good job
    </yann>
    True, I've found this also. But it's always important to check the license. Many commercial products specifically forbid any form of decompilation, ruling out this approach.
  81. OSS vs. Commercial[ Go to top ]

    Rod,

    I know that. But it's just a disclaimer in case some real issue happens with decompiled code (such as writing cracks or altering software). Using decompilers is factually tolerated for debugging issues and of course not advertised. It's one of those things everybody does but dares not saying. I've already told support from commercial companies about bugs/fixes I found after decompiling classes. Some companies are loose with that. Others though are extremely touchy about this but I won't give names. :)

    So I'll reformulate: decompile only if case of emergency (support is too slow, blocking problem, ...), fix if you can without altering the commercial packages and don't tell how the problem was fixed. :)

                    Yann
  82. Decompiling commercial software[ Go to top ]

    "But it's just a disclaimer in case some real issue happens with decompiled code (such as writing cracks or altering software)."

    I always thought it was to protect the intellectual property of the company. Afterall, isn't the source code all the diffentiates one app server from the next? They probably don't want people poking their noses around to see what clever code they have for caching, optimizing, etc.

    Ryan
  83. Decompiling commercial software[ Go to top ]

    Ryan,

    "They probably don't want people poking their noses around to see what clever code they have for caching, optimizing, etc."

    Obviously not :) This is why some some people use obfuscators. You're right, it's much more an issue with intellectual property. I was focused on smaller scale issues with decompiled code.

                    Yann
  84. OSS vs. Commercial[ Go to top ]

    Yann,

    I agree with what you say sometimes we need to bend the rules, but at what cost ?

    Why cannot we have an independent body that can certify an application or product. Why should we open up the source code ?
    Why should every developer be involved in it ?
    For example, when we buy a car, we know that it has been certified for safety. We don't rip the car open to see for ourselves.
    Please don't get me wrong, I like to tools I use, but I don't see how programmers employed full-time can manage to study the source code of OS tools that they are using.
    I just don't get the time.
    I would be interested to know how often programmers open the source code at all.
    I have found more answers to my questions or code snippets on Google groups than open source.
    Why cannot we use such forums as a means of sharing source code ?

    Also i don't believe the notion that using open source would mean decreased job cuts. I have not seen a single company citing hardware or software expenditure as the reason behind job cuts.
  85. Checking the source code[ Go to top ]

    Checking the source code is a little extreme case, at least unless you are very familiar with particular OS component. It all depends on circumstances, most of the readers of this thread would probably have no problems tweaking the code of Jakarta Commons but would probably pass the C code of MySQL or Linux.

    Anyway Open Source still means that code is "out in the open" and the potential deficiencies are discussed and easy to identify before you make the commitment (see above discussion about JBoss mailing lists as example).

    Open source also means that if you buy support (MySQL.com, RedHat), they have to be a little more competent to impress you and make you want to spend a money. After all you can get code snippets yourself and lookup the bug reports yourself, so they have to try to go beyond that.

    I think whole discussion about JBoss === the OSS is a little misguided. Unlike MySQL or RedHat, JBoss still seems to look for the business model, certainly charging for documentation is not the best marketing ploy, unless of course, you actually believe that there are 2 millions of new production installations yearly, each in the need of printed documentation ;-).

    Michael
  86. What problems have you had with Lucene?[ Go to top ]

    I just implemented Lucene in my current project and it went great. While I haven't used commercial search engines (save Microsoft Index Service) to compare against, I found Lucene easily customizable and very fast. My application dealt with military names and nomenclatures and Lucene let me write and plug in custom tokenizers and indexers to fit my application.

    Could you be more specific as to your problems with Lucene?
  87. Lucene[ Go to top ]

    What a coincidence.
  88. Lucene[ Go to top ]

    DSpace
     from MIT uses Lucene. They don't seem to be too stupid. :)
  89. Lucene[ Go to top ]

    Guys, there is a separate lucene post, please discuss lucene there.
  90. Lucene[ Go to top ]

    Floyd,

    I'm afraid we started that Lucene thread before it was coincidentally posted in the news. But I'm done with it for now. :)

                    Yann
  91. More about OSS and less Lucene[ Go to top ]

    Sorry Floyd, but the posts were more about OSS and the quality of it and less about the actual usage of Lucene. The discussion of it in this manner follows this thread better.
  92. I think to much of the discussion around Open Source revolves around price, doing right thing, etc, wheras OS can quite often mean better quality.

    From my perspective the quality is how well product meet the promises/expectations, OS definitelly has advantage here.

    With OS components:
    - very few OS projects oversell their products, on the contrary most underdocument the features,

    - people are more free to criticise and comment the OS products, there is no fear of Legal Dept scanning the mailing lists, thus much better is available before you make a commitment,

    - bug reports are available to everyone, it does not take 3 days to confirm that your problem is a known bug,

    - time to get the response from mailing list is comparable with response from paid support, and usually beats the support email "Issue numer #143256 has been logged for your problem",

    - worse come to worse you can patch the product yourself, and the developers would often provide a guidance on how to do this,


    Well, here is my experience with commercial support:

    - 30% of the features demonstrated in samples and tutorials, are only usable in samples in tutorials,

    - you never know what works and what does not until you used 80% of your budget and the final product test is just weeks away,

    - 3 days after logging the issue you get the call "It is a known issue it will be fixed in release 4.0 scheduled next spring",
    ... needles to say, well past your deadline,

    - no you can not fix it youself, even if you know how - you can get to jail for trying to do that,

    - and last but not least Vendor ping-pong: Vendor1: 'its Vendor2 issue', Vendor2: 'no problem is with Vendor1 ', ...

    There are very good commericial products too, but very many cost you much more then just licence price.

    Michael
  93. Cost benefits of using J2EE and Open Source[ Go to top ]

    We are under NDA so I will not divulge my company other than to say that we are doing a final beta rollout of an enterprise class software based on Linux +Jboss 3x + SAPDB + SQL Explorer from Sourceforge+ (Netbeans/ VI ) + Apache XML and Jakarta Commons + JDom + DOMAPI (for javascripts)+Xdoclet. In fact the only software that we have bought are JProbe Suite, InstallShield and some tools for Tech Writers to help in their documentation.

    We have saved a bundle and whats more, we have developed this attitude of trying out open source componenents and even fixing them to do what we want , rather than creating any component from scratch. This has saved us a lot of time while interacting with other Open source developers has imporved our coding standards.

    We are sanguine about our future as our closest competitors are running on WLS/Websphere and Oracle/Sybase/DB2. So evenwhile feature wise we are the same, TCO for our product is a fraction of what it is for those guys.

    We could not afford the 10k+ paid support from Jboss Group and so getting support from Jboss is a pain but bearable. SAPDB is flawless. Jakarta components are very well documented and Xdoclet gurus are very responsive.

    So far , its been a great ride !
    Cheers!

    dd
  94. "We could not afford the 10k+ paid support from Jboss Group and so getting support from Jboss is a pain but bearable."

    <vendor>
    Pramati premium support costs only 25% of license fee (2cpu Standard Edition license costs $5000, list price. Support for this would be $1250.)support costs
    </vendor>

    Cheers,
    Ramesh
  95. First this remark : I believe that the quality of the code in OSS is on average higher then commercial software code. OSS is written by motivated developers with a vision, while the in commercial project teams programmers tend to be less motivated and less capable.

    Second : I want to promote a new motivation for developing open source : Companies should develop all software that's not their core business in open source. That allows companies to share development efforts on software that is not their core business.

    Third, OSS documentation on the other hand seems to be less elaborated then the commercial counterparts. But it seems that it didn't keep a lot of people from using software like JBoss, Ant, Struts, PostgreSQL, MySQL, ...

    Fourth, Another new and exciting OSS project which I would like to mention here is jBpm.org. The first release is expected one of the next days. It takes business process models as input and manages the user-interface for executing the processes.
  96. Open source support -- Irony[ Go to top ]

    Tt is ironic that so many detractors of open source say that you shouldn't use open source because you can't get support. Then when a group like JBoss starts offering support, everyone gets in arms that it's not true open source anymore.

    Seems to me like the JBoss group saw a need, and filled it.

    Personally, my experience getting help from open-source mailing lists has been almost uniformly better than getting help from commercial vendors where you often have to pay to open a ticket, then send them the source, them wait for someone to get around to figuring out..etc,etc. This hasn't always been true. Macromedia's support for JRun (which we do use) has been excellent.

    I think a good idea for a services company would be to offer support for a fleet of existing open source apps, like most of the stuff that comes out of xml.apache.org, jakarta.apache.org.

    IBM saw a trend in services, and is making good money off it (something like 40% of revenues, with pretty high gross margins I'm betting).

    And, what it really comes down to is business: Does it make technical sense to use OSS? Is the ability to see and debug valuable to me, or will my group/developers just find workarounds? Is it more efficient in that case for me just to buy Commercial, which will maybe save me money in other areas? To me the critical factors consistently are ROI, and technical features and feasability. I may use a commercial app-server, but I use tons of open-source, Castor, Struts, OJB, and a whole ton of other projects. They just make sense for where I am (research).

    Jason McKerr
    Northwest Alliance for Computational Science and Engineering
  97. It is useless to debate whenever OSS is good or bad. If it’s good enough - it will be used, then the marked for that kind of application is destroyed.

    But don't make the mistake to believe that the software is put together only by idealist around the word aka "The Cathedral and the Bazaar" though. Without IBM for example, Linux would have gotten nowhere.

    OSS is the last hope for the companies that have lost the lead: "As we can not compete, let’s gather together and destroy this market".

    IBM (and a bunch of other companies) tries to destroy the market for Windows with OSS/Linux. Microsoft likewise destroys the market for Java by giving .NET to OSS/Mono.

    Illusions anyone?

    Regards
    Rolf Tollerud
  98. What?[ Go to top ]

    Rolf,

    "I am constantly amazed to why Open source un-American movement gain so much space."

    "The open source people can only make software by stealing."

    Care to defend these statements you made?

    Ryan
  99. About Open Source etc[ Go to top ]

    Ryan,

    I made that statement a time ago when I was a little irritated. I am somewhat more positive to OSS now with Ximian/Jakarta and all. But I think you can not deny that it works best when they are copying something or for fixing bugs. I am just striving to position everything in perspective - OSS is a little over hyped. They can not cannibalize old Unix apps forever.

    The Royal Bank of Scotland. I am sure that I am not the only one that is interested in this application. Is there any link? Where is more information? BTW the Gartner report that I read about was not from 1994 but from 2002.

    Anyway I am only interested to get everything in proportion. For example, Microsoft is treated extremely unfair in this forum IMO. Microsoft is not guilty of criminal behavior and the Department of Justice has not formally alleged that Microsoft is breaking any antitrust law. Instead, Justice claims that Microsoft is violating its 1995 settlement of an earlier, untried antitrust suit.

    Moreover, In Pricewaterhouse Coopers survey, published today, based on 1000 CEO's opinions in 20 countries, Bill Gates is ranked first as an company leader - and Microsoft as the most respected company (together with General Electric). This is in addition to the former survey, which showed Microsoft as "Most Desired IT Employer of 2002". So how goes this together with the constant bashing of Microsoft in this forum?

    TSS is very enjoyable reading (how often is it possible to combine business with pleasure?) But it is unfit for the forum to listen to hype from *any* sources. SUN, Oracle, IBM, Microsoft, all have marketing departments, but 200.000 experienced consultants should not be swayed by anything but facts. This is not Slashdot.

    So I would like the forum to become more objective. But if this is not going to happen, if MS all times is presented as "good for only small solutions or desktop applications", then I will go on as the "court jester" (in the style of Alexander Dumas).

    Regards
    Rolf Tollerud
    Just a little "childish garbage"
  100. About Open Source etc[ Go to top ]

    Rolf,

    "So I would like the forum to become more objective. But if this is not going to happen, if MS all times is presented as "good for only small solutions or desktop applications", then I will go on as the "court jester" (in the style of Alexander Dumas)."

    You seem to be keen on Gartner reports. Well, you should read "J2EE, .NET and Web Services: Software Architectures for the New Enterprise" (Gartner Nov. 2002). They are the ones who say 'The War of Architectures Is a Border Dispute' and that 'Through 2006, more than 75 percent of new application projects for more than 1,000 concurrent users will not consider .NET as their potential platform; more than 75 percent of new application projects for fewer than 100 users will not consider J2EE (0.7 probability).'

    I don't think Gartner reports are necessarily the best source either when it comes to having an objective view of the market. They often contradict themselves, depending upon who put pressure to have a report. But since you mentioned Gartner...

                    Yann
  101. About Gartner reports[ Go to top ]

    Yann,

    "You seem to be keen on Gartner reports"

    Please, I referred to Gartner only once!

    But I cannot contradict you, as Gartner seems to have an obsession for the number 80%. I tried to find the report by searching in Google, and I got hundreds of hits, some follows. A search on Google on gartner 80% java:

    "According to a recent Gartner survey, 80% of US online consumers are aware of person-to-person (P2P) e-payment services"

    "Gartner Predicts 80 Percent of External Storage Will Be Networked by 2005"

    "Gartner research shows 80 percent of mission-critical application service downtime is directly caused by people or process failures"

    "Gartner, 80% of all enterprises use the Java language"

    "Gartner analysts about their predictions ... Which is going to win - Java or .NET? ..Together, they'll have about 80 per cent of the"

    "Gartner, By 2003, approximately 80 percent of all platform vendors will support Web services architectures"

    "practice for enterprise software engineering, Source: Gartner. 20% Automated / 80% Manual"

    "Gartner Inc., a mar- ket research firm, said Java ... predicted that at least 80 percent of mobile phones will support Java by 2006"

    "Gartner, 80 percent of all Java deployments overpay.."

    and so on and so on..
    Maybe we are in the wrong field? The Gartner business seems very easy and profitable..
    But, we both now that large project usually fails, no survey needed.

    Regards
    Rolf Tollerud
  102. About Gartner reports[ Go to top ]

    Rolf,

    "But I cannot contradict you"

    You don't need to! :) Another report from Butler ("Application Servers - The Facts and How They Compare")states that:

    - .NET lacks credibility at the enterprise level;
    - Rate scalability 4/10 (2nd worse mark out of 10 application servers surveyed);
    - Rate integration 3/10 (2nd worse mark);
    - Rate security 5/10 (2nd worse mark);
    - Rate market strategy 8/10 (best mark);
    - It's unlikely that organisations with heavy commitments to J2EE will migrate to the .NET platform.

    Of course, they also say very good things about .NET notably regarding web services but on the other hand, they say that web services are not a mature technology. Given that most enterprise needs of big projects these days concern integration and scalability, .NET is not exactly well positioned on non-Microsoft platforms. Again, this is not the J2EE community saying that but one of the most influent trend analysts.

    "Maybe we are in the wrong field? The Gartner business seems very easy and profitable.. But, we both know that large project usually fail, no survey needed."

    I don't think Gartner business is a very easy one. It is not so easy to follow the market trends and the people they contact to collect data to build statistics sometimes have an interest in that the outcome of the reports does not contradict their own enterprise IT strategy, etc. It is certainly profitable though.

    The problem I see with Gartner (or Forrester, Butler, ...) reports is that whether the contents reflect reality or not, they are an important lever for many decision makers. It's a bit like financial analysts in the stock market: they can influence the market. So if today a Gartner report states that JDO is not going to be adopted with a 80% probability, what do you think big accounts will do ? And how will they react to the Butler report that says that .NET will likely not take off because it lacks integration and scalability ? Decision making is all about risk analysis. Of course there are other levers and not all decision makers take those reports for granted.

    That said, I'm pretty certain you will find Gartner reports that say just the opposite. :)

    Now regarding your statement that large projects usually fail, I think it would be closer to reality to say that large projects, due to their complexity, the likelihood of political issues and inherent demanding communication requirements, have a greater propension to fail than small projects. But I guess there will never be any solution to those problems.

                    Yann
  103. About Gartner, Butler etc reports[ Go to top ]

    Yann,

    OK, you made your point. I promise I never refer to Gartner again. And to think that people actually pays money for these reports!

    Regards
    Rolf Tollerud
  104. About Gartner, Butler etc reports[ Go to top ]

    Rolf,

    "OK, you made your point. I promise I never refer to Gartner again. And to think that people actually pays money for these reports!"

    They are not that bad either. I usually find them very informative, but one needs to take them with a pinch of salt (subjectivity shields on) as any other kind of precompiled "reader's digest" information, that's all.

                    Yann
  105. About Open Source etc[ Go to top ]

    I leading a re-engineering effort for a major Insurance company. They have a renagade group in IT that is offering an alternative to their standard WebSphere deployment environment, which by the way, has hamstrung project productivty through operations hell, which is designed to make the admins live's easier.

    The new deployment platform which includes Tomcat, Apache and Linux, it's targeted towards enterprise applications that diminimus SLA profiles (although we could easily handle the demands of high volumes). We use Struts for the presentation tier and JavaSpaces as the middleware plumbing, which provides the QoS we need for our mission critical application.

    Our development time has been reduced, our architecture is simplified (much more so than if we used EJBs), it's extremely robust, scalable, fault tolerant and highly available. we put servlet sessions in a shared JavaSpace to enable clustering and auto failover.

    The passive collaborative paradigm of JavaSpaces allows us to create completely decoupled services that are easier to build and maintain. It was a breeze to enact complex workflows that are persistent even if a server goes down.

    -ernie
  106. About Open Source etc[ Go to top ]

    Ernie,
      You're using Javaspaces? Cool. Could you provide more info? I've been trying to push the use it to solve some of our issues but need some real life expamples to make the point.

    Mark
  107. About Open Source etc[ Go to top ]

    Mark,

    Jini and JavaSpaces are being used in many high profile, high demand situations. Notable telecoms including Erricson, Nokia, and Cisco for telecom switches. The military uses it for battle theater management, unamed aircraft manufacturers are using it for grid style computing to solve fatigue analysis applications. The new JRun server uses Jini for dynamic deployment of web components, etc, etc.

    We are using it because we can get the same or better QoS of expensive clustered app servers using open source solutions while avoiding the complexities of EJBs.

    -ernie
  108. About Open Source etc (Jini/Javaspaces)[ Go to top ]

    I know it is being used other places but I usually just get generic descriptions. I was hoping to use it solve some of our cluster issues too. Could you describe how you use it in a little more detail? I think this type of solution is overlooked way too many times.

    Thanks,
    Mark
  109. About Open Source etc (Jini/Javaspaces)[ Go to top ]

    We use GigaSpaces, a commercial implemntation of JavaSpaces. It comes with a utility called SessionSpace that puts sessions in a space that can be shared by multiple Servlet containers, thereby providing clustering to the servlet containers.

    I have done similar things in the past with a custom implementation using the Java Shared Data Toolkit (JSDT), a similar technology to JavaSpaces, unfortunately no longer supported by Sun, even though it's listed as a standard extension on java.sun.com.

    If you're looking for details on how to extend the session mechanism and distribute it across multiple JVMs, there are a couple of articles on the subject on www.javaworld.com under the following titles:

    Manage Distributed Sessions
      by Kelly Davis & Robert DiMarco

    Take Control of the Servlet Environment
      byThomas E. Davis and Craig Walker

    The solution we use to decouple the presentation tier from the business tier is to use a business delegate that is responsible for putting request entries into the space. These request entries have a value object that contains the request parameters, a contextural command and UID. Then the deligate registers a notification with the space that will fire an event when a response entry is put into the space that has a UID that matches the request entry's UID.

    There are various business components, implemented as Jini services, that listen for the request entry with a specific command value or request subtype, they take the request entry, process the command, then deposit a response entry that contains the results payload. The business delegate is notified that the processing is complete, takes the response entry and returns to the client.

    That was the synchronous case. In an asynchronous case, such as background processing, we don't register a response listener in the space, we just put the request entry in the space and immediately return to the client.

    We have implemented various generic business components that interact with the space, some that specialize in managing specific workflows, others that process/record business events, a VO mapper, messaging manager, etc.

    -ernie
  110. About Open Source etc (Jini/Javaspaces)[ Go to top ]

    Ernie,
      Thanks. I wondered if you were using Gigaspaces. I've looked at it and have talked to them.

    Mark
  111. About Open Source etc[ Go to top ]

    Rolf,

    You are a partisan, admit it. Just as those, like myself and most others on this forum, are Java partisan. Please don't insult us by claiming the high ground of objectivity. It simply isn't true.

    Your arguments are strinkingly similar to the tautology of politicians. You assemble a number of anecdotes that support your point of view in order to raise the ante of your facade. Then you envoke the principle of propoganda, if you say it enough times, it shall be true.

    The plain fact of the matter is that Java and Microsoft based solutions are philosophically opposed. Java is a best of breed approach and .Net is a one stop shop. It just so happens that most enterprises cannot risk a single point of failure, therefore Java is the superior choice.

    -ernie
  112. About Open Source etc[ Go to top ]

    Rolf,

    Whose Alexander Dumas? Iz he workin for Microsoft? never herd of this guy(unless you ment Alexandre Dumas.
  113. Idealists?[ Go to top ]

    Rolf - "But don't make the mistake to believe that the software is put together only by idealist around the word aka "The Cathedral and the Bazaar" though."

    Who said OSS is put together by ideasists? Did somebody in this thread say that? Do you think that was what Raymond was saying is his book? Have you read the book? Have you ever talked to him? I don't think that was his point at all. So where did you get the impression that somebody thought that OSS was developed by idealists? Or is that something you just thought YOU would throw out there?

    Ryan