IBM Disputes BEA's Claim of Big Gains at WebSphere's Expense


News: IBM Disputes BEA's Claim of Big Gains at WebSphere's Expense

  1. IBM is disputing recent claims by BEA that BEA is supplanting IBM in application server deals and beating it in head-to-head competitions. IBM makes the point that BEA is counting "project wins" rather than overall revenue, particularly since BEA's software licensing proceeds reportedly took a 28% plunge. IBM still sticks to measuring growth in terms of revenue.

    Read the whole article by Darryl K. Taft on,3959,476252,00.asp.

    Do you have opinion on why IBM is still growing its revenue in application server field while BEA is not?

    Also, BEA being the provider of "the only nonproprietary solution" (Alfred Chuang, BEA CEO), that is the only major vendor that is not trying to also peddle its hardware, DBMS, etc. -- is it an advantage or a handicap?

    Threaded Messages (32)

  2. I think marketshare measured in terms of deployments is a better measure than marketshare in terms of revenue. If company A with a zealous project manager spends $500K on a server "A", but 10 other companies spend $50K each on server B, shouldn't server B be recognized as having better marketshare?

    These revenue based marketshare reports are a major reason why free servers like jboss and tomcat get less respect among executives, they usually don't even make it into the analyst reports!

    We need a proper marketshare survey. TSS will organize such a survey later in the year.

  3. I agree Floyd. Both perspectives are important, but that really depends on who yuou are. As a sharleholder in a company, I don't much care about numerical deployments.

    However, as a statement of "market share" I feel the actual number of deployments is far more important. If we stuck with revenue, JBoss and Jonas would have no market-share.

    Theserverside a while back posted a link to an article regarding some Giga (or maybe on of the other research firms) study on marketshare that was all reveneue based. I emailed the head of the study asking about that. He basically said that the financial incentives for his firm to do the study was to sell it to software companies. So of course their gonna look at exactly what these comapnies want them to.

    Because of this, I think there needs to be some incentive (or some independantly funded research study) to study the "real" numbers. Knowing for sure JBoss's real deployment numbers in terms of market share percentage would be great for people trying to get it in the door.

  4. I agree that we need to take the number of deployments into account. This, however, opens up another issue: does the "size" of a deployment matter?

    For example, does a deployment running a major e-commerce site count as much as another deployment running internally in a 8-person department? If not, how do we quantify it?

  5. <quote>
    For example, does a deployment running a major e-commerce site count as much as another deployment running internally in a 8-person department? If not, how do we quantify it?

    IMO, counting "deployed CPUs" should give a more accurate picture of each app server market share.

    -- Igor
  6. Just an example of the relationship between figures and reality:

    My company pre purchased a large amount of Websphere production CPU's.

    This was over 2 years ago as part of a very large deal with IBM.

    We have so far deployed around 20%, mostly Web container apps. ( JSP/Servlets )

    Most 'new' J2EE apps utilising the EJB container, are being developed using WebLogic 6.1 - 7.0

    Needless to say, we will not be buying another large amount of Websphere licenses next year.

  7. It seems the WAS is popular among managers. They only know that it is from IBM. OTOH, WLS is favored by developers. It is my impression from reading the various TSS posts. So don't flame me for that!
  8. Do you have opinion on why IBM is still growing its revenue in application server field while BEA is not?
    Do you have an opinion on why a SCSI hard drive from IBM costs 10 time more for an AS/400 then the exact same drive for one of their High End Intel servers with the same level service agreement?

    How about RAM? Do you have an opinion on why 128MB of IBM AS/400 RAM goes on ebay for $650? That's EBay for God's sake! What do they want for that puppy new?

    I think IBM just has a whole lot of customers who are clueless and will buy anything big blue says is good. Happens here in the AS/400 side of the shop all the time. IBM has achieved customer lock in, and the customer has to pay whatever IBM demands in order to stay afloat. To make matters worse, the customer base usually only knows about IBM products and solutions.

  9. IMHO, revenue, especially with IBM, and perhaps even market share in general, does not have real meaning. If a company decides to use IBM products, in most cases the reasons won't be technical at all. It'll mostly be a strategic decision - i.e. choose a company that offers everything, from hardware and operating systems to a database, app server and IDE. Of course this is both an advantage and a disadvantage.
  10. I can the next couple of news items already.

    BEA disputes IBM's dispute that BEA's Claim of Big Gains at WebSphere's Expense

    followed by

    IBM disputes BEA's dispute that IBM Disputes BEA's Claim of Big Gains at WebSphere's Expense
  11. Peter,

    "I think IBM just has a whole lot of customers who are clueless and will buy anything big blue says is good."

    Well said... This statement just about sums it up.

    WebLogic does not compete with WebSphere, it competes with IBM.

  12. I think there is another big thing IBM has going for them: many shops adopting J2EE today are larger shops parting with their mainframes.

    Since the mainframes are mostly IBMs,

    a. There is an existing relationship with IBM reps.
    b. Management may want to leverage existing skills, such as DB2 skills (beneficial to IBM's end-to-end solution offering).

    As a result, IBM/WebSphere often has a first shot at mainframe-turn-J2EE customers. Besides, no one ever got fired for buying IBM.

    It would be interesting to see how BEA and IBM fare in shops with no previous presence by either vendor.


  13. Well Boris, That's exactly the point ! BEA is no IBM in size or in tactical advantage yet BEA still beats IBM head to head when it comes to choosing a reliable app server. BEA only wins by selling facts about thier products while IBM pulls all sorts of tricks from thier hats to win, thats if they do win. + IBM renamed most of its products to WebSphere something, like MQ for example. I bleive Every sale of WebSphere MQ gets reported as a websphere sale, which could say alot about the numbers IBM is marketing with :-).
  14. I can't believe I agree with most of the posts so far:
    IBM has a lot of leverage
    BEA has a much better
    As naive users/mgrs become experienced users/mgrs
    J2EE applications approach WebLogic

    To believe IBM's claims you would have to believe customers are just staying naive (this may certainly be the case for some but not most.)

  15. One of my previous clients we decide to use Weblogic and IBM gave away WebSphere server, Studio, VAJ for free if we buy their AIX hardware, guess what Mgr. decides to go with IBM products. I don’t how IBM reports this sale?

    - monic
  16. "Do you have opinion on why IBM is still growing its revenue in application server field while BEA is not? "

    Could it be because part of every sale of an AS/400 (iSeries) is booked as license revenue for WebSphere, since it's bundled? I'm sure there are plenty of other product bundles that IBM sells which include Websphere, which may not be actually wanted or used, but are included anyway.

    Mind you, I don't know they do their accounting this way, but I wouldn't be surprised. Plenty of other companies do this, but IBM has the deepest, widest stack, so they have the most opportunities.
  17. IBM is growing WebSphere revenues so that it can show near parity with (or even a lead over) WebLogic so that developers will support WebSphere too.

    Frankly, it is very good marketing. WebSphere was hopelessly behind before. Now it is a de facto "must also support" for many ISVs.

    OTOH it is costing IBM hugely. They are paying some customers to take WebSphere (you buy WebSphere for $X and we'll give you a discount of 1.2 times $X on product Y.)

    But if it buys them into the "next big thing", then maybe it's a good investment. BEA doesn't have the same leverage, so from that point of view it's a bit unfair. However, in the trenches BEA still has a significant lead over WebSphere.

    The good news is that both WebSphere and WebLogic (as well as other J2EE app servers like Oracle) are improving rapidly because of the competition. Some are obviously "dropping out" of the race (e.g. HP), and JBoss is wiping the decks (is that the right phrase?) with whatever is left over, making sure that .NET has no wiggle room at all to compete with J2EE.


    Cameron Purdy
    Tangosol, Inc.
  18. All the comments in this thread state that no one chooses WebSphere because it is technically good only for IT political reasons. These are all opinions and not true.

    Honestly speaking, Any technically sound architect or developers should be able to succeed with WebSphere, WebLogic, Oracle, or what ever App Server. They should be able to develop J2EE applications that perform well, scales well, etc... If anyone claims failure with one App Server and success with another either doesn't know what they are doing or they just lie to their superiors because they like an App Server.

    Thoughts my own and not IBM
  19. I agree with Roland. Surely the purchase of the Application Server software by whatever vendor is the final step in development, not the beginning?

    I have used both WAS and WLS (plus most of the others!) and have had success with both. In terms of scalability, there are occasions where WAS has the advantage, but we are talking very large systems. WLS is favoured by developers because it is elegant and you can get up and running quickly (something that IBM is trying to remedy). I know of at least 2 sites where development is done on WLS and deployed to WAS for production.

    The fundamental principle, however, is that architecture should be designed around the J2EE principles. Developers then have to work around the nuances particular to the App Server that will be deployed in production.

    I am sure that this BEA vs. IBM debate will continue for some time to come...
  20. I worked at a very large government organisation in Australia.

    They had a deal with IBM, where IBM would give them a free license to use any IBM software, and only after 12 months would the organisation start paying license fees.

    This was an innovative approach to licensing whereby the organisation was able to experiment with many products (including Lotus products BTW), Websphere among them. Since these guys had the opportunity to evaluate a top-end appserver, they wound up being a Websphere shop as far as J2EE appservers were concerned.

    If BEA was as creative in it's approach and looked at the longer term rathern than year-end revenue, maybe they could gain more revenue in the long run.

  21. From the app-server-market point of view it's easy to see number of installed CPUs is the measure of success.

    However from the point of view of the company *shareholders* the revenue (well, profits on revenues) is the important data point.

    I'd like to see IBM explain how it proportions 'bundles' of softwares or hardware/software to its various revenues.

    On the other hand, it's no point selling 1,000,000 units if you only make $1 per unit, if you could sell 100,000 and make $100 per unit. Volume businesses often miss out of niche opportunities. (I don't know how BEA fits into this scheme though).


  22. scot mcphee wrote:

    "I'd like to see IBM explain how it proportions 'bundles' of softwares or hardware/software to its various revenues."

    The picture that I'm beginning to get is that IBM may be bundling Websphere with hardware sales at little or no cost to the customer whether they need it immediately or not. This would make sense of the data BEA is reporting about Weblogic 'replacing' Websphere in a number of places. But it would also mean that IBM has been cooking their 'market share' figures.

    Any comments from someone (IBM employee or customer) with better data than I have?

  23. Websphere and WLS's days are numbered. Go check out the download statistics for JBoss at and see why this bickering between IBM and BEA is meaningless.

    We have 239K downloads this month and have had 1.8 million downloads since March 2001. I remember a recent news article from Sun talking about how the Sun J2EE reference implementation had reach 2 million downloads for its lifetime. We have reached the milestone in only 16 months.

    Shortly JBoss will be posting case studies of companies large and small that are using JBoss and JBossGroup services(docs, consulting, training, and support) to build successful cheap J2EE based applications.
  24. It was somehow odd to see words "cheap" and "J2EE" in the same sentence. Numerous case studies suggest that companies adopting J2EE long for scalability, availability, and the rest of the "bilities" rather than for the cost containment. lists their monthly downloads rate as 150,000+ plus. 150K or 239K, there is still a wide gulf between a download and a deployment. How many of 2M downloads of Sun's reference implementation are actually deployed? (Yes, I do appreciate the difference between JBoss and the ref. implementation :))

    Also, how does outsourcing "docs, consulting, training, and support" to JBoss Group or any other consultant make anything cheap?

  25. Hi Bill,

    You will always have a different measurement of success than the WebLogics and WebSpheres of the world. They could in turn say that you are a complete failure because your revenues from licensing are $0. They would be wrong to say that JBoss is a failure, but (perhaps those multi-colored pills are getting to you) you are silly, and frankly your value system is FUBAR, to gleefully claim that their days are numbered.


    Cameron Purdy
    Tangosol, Inc.
  26. After putting our ecomm website on Weblgic 7.0 it has been a nitemare for us to keep the site up. Weblogic 7.0 is still a hype where as Websphere 4.03 runs 30% faster than Weblogic 7.0 and thousnad times more stable we had zero down time with Websphere app server but with Weblogic we go down every 2 days.
  27. I believe much of the debate comes down to credibility and who you can trust.

    Did BEA say they never lost? No. Did BEA say they beat IBM in all 1000 transactions? No. BEA reported that they did lose, but not very often to their primary competitor and in a number of cases they replaced WebSphere. I don't see anything in this thread that would discount statements as such.

    We all know that IBM likes figures, however, they did not respond with similar metrics because they likely do not have any. Instead they tried to change the game by talking about market share.

    Ironically, Billy Newport (IBM developer) just stated in another thread that WebSphere 5.0 is coming in November 2002, but Scott Hebner, WebSphere Product Manager, states in this very rebuttal that WebSphere 5.0 is due out next month (September). Billy went on to discuss IBM's WAS 5.0 Enterprise offering but failed to mention that it will not be available in November. Nor did he mention when it would be available.

    IBM may be regarded as a good marketing company, but this rebuttal just raises another one of IBM's questionable practices of not breaking out WebSphere application server revenue so that we're all left guessing.


    Mike M.
  28. Now wait a minute,
    Scott is big enough to speak his own mind but you're misstating the story. Scott didn't say September, that was the authors opinion.

    "Hebner said the disruption made by the close of WebGain Inc.—which made tools for WebLogic developers—and the sale of its technology to Togethersoft could create problems for BEA in its developer community. He also said WebSphere Version 5, which is due out next month, will enable developers to build and enable WebLogic applications in an IBM environment. "

    The text "which is due out next month" is the article authors opinion of when the product will ship and obviously, the authors opinion is just that, an opinion and it happens to be wrong in this instance. It's not a quote from Scott, so no irony required.

    Enterprise edition will ship very shortly after the base product ships. We're in the final stages now of prepping the code for the final beta in September.

  29. Billy,

    Let's stick to credibility.

    IBM released a "beta" in December of 2001 and subsequently touted they were first to J2EE 1.3. Now, IBM is in the unenviable position of being last to J2EE 1.3 and ironically, may not ship a J2EE 1.3 certified version of WebSphere 5.0 until 2003.

    The truth is that the November release of WebSphere 5.0 will likely only include Express or Developer edition which will only support non-clustered, Servlet/JSP applications and will not deliver on production J2EE 1.3 application deployment.

    IBM chooses to gloss over this fact in all of their discussions with the press and although IBM has known for some time that their announced WAS 5.0 release date would slip, it took a fair amount of prodding until IBM would finally admit (with plenty of spin) that WAS 5.0 would slip.

    IBM Nudges Back Release of WebSphere 5.0
    James Niccolai
    August 23, 2002

    Frankly, I don't see the connection between Studio & WAS 5.0, but the article still implies that J2EE 1.3 is as close as November. The truth is that IBM has yet to announce when they will release a product that is comparable to BEA's WLS 7.0. Instead, IBM will make a big splash in November (or December) and when Enterprise Edition finally arrives sometime in 2003, new IBM customers will be able to begin testing and deploying J2EE 1.3 into production - and existing customers can start planning their porting efforts from WAS 3.5/4.0.

    Of course, much of the WebSphere product line requires WebSphere Application Server. So the obvious question is when will IBM begin revising release dates on Portal, CrossWorlds ICS, and the other 30-40 products that must be ported to the new release of WebSphere Application Server.

    It looks like IBM will leave us guessing.

    BTW, I certainly don't agree with your view that the press simply manufactures their information based on "opinion".


    Mike M.
  30. Mike,
    "BTW, I certainly don't agree with your view that the press simply manufactures their information based on "opinion"."

    Of course, it's not manufactured, the author had out of date information, no fault of his own. Rumour is probably the wrong choice of words looking back at my posting, however, I've never heard Sept as even a possible GA date for WAS. What he said in the article wasn't why I responded at this late stage in the thread.

    I responded because you tried to use a quote from my-self and a quote incorrectly attributed by you to Scott to discredit him to readers of this thread. Something I noticed you didn't apologise for or even acknowledge as a inaccurate in your follow up.

    Studio, or better stated WSAD, needs a single server version of WAS to implement the integrated 1.3 test environment and there is the connection.

  31. Mike,

    I certainly don't think they manufacture information, but have you ever been interviewed by some of these press-comapnies like infoworld or whatever? Mostly they are asking you questions and writing your answer down as you go along. You can literally see (or hear, if it's a phone interview) them losing phrases, words. They annotate and paraphrase. Then when they go back and look at what they wrote in their notes, they forget the context that they wrote it in.

    Invariably things get butchered. It's happened to me, and everyone else I've ever met that's ever been interviewed by _any_ press body.

  32. Billy,

    Since you were unaware that IBM had come out in the press with release dates for WAS 5.0, I went back to get you a few examples.

    Here's Scott quoted as saying "WebSphere 5.0, which IBM expects to release the middle of the year, Hebner said".

    J2EE Delay Won't Impact WebSphere Releases
    Rita-Lyn Sanders
    e-Pro Magazine
    April 02, 2002

    Here are some quotes from Steve Mills and John Swainson with an availability of WAS 5.0 in June.

    IBM Unveils New WebSphere Products
    Elizabeth Montalbano
    May 08, 2002

    Here's John Swainson again with shipment of WAS 5.0 in the Summer and packaged versions to hit the market in July.

    IBM Sticks another Candle in WebSphere’s Icing with V5
    by Rita-Lyn Sanders
    News Editor
    May 9, 2002

    Under the credibility microscope, I would respond to Cameron by stating that IBM has done an excellent job of "marketing" expectations.

    I don't know why you would expect me to apologize for holding these individuals accountable for what you term as "out of date information". Neither IBM nor any of these individuals ever attempted to set the record straight so they are to be held accountable for misleading us.


    Mike M.
  33. Mike,

    Although IBM WS 5.0 is much later than BEA WL 7.0, it is a widely held opinion that WebLogic 7.0 could have used a little more time to finish baking itself. I still think that WebSphere is at least six months to a year behind WebLogic, but reliability/availability are king for commercial GA releases. I know these companies are caught between a rock and a hard place -- damned if they do and damned if they don't. When a vendor waits to long to release software or access to new features (e.g. J2EE 1.3), they are criticized for being behind, and if they release it too fast, they are criticized for the quality of the release. I think IBM did the right thing by releasing the 5.0 "beta", but they did a pretty poor job of managing expectations set by such an early "beta".


    Cameron Purdy
    Tangosol, Inc.