"There is simply too much complexity, both initially and in maintenance over time, to retrofit this market. That portends, in time, yet another tornado as the whole world swaps out client/server infrastructure for web-enabled."
However, this is, by and large not what is happening. Companies are not redeveloping all their applications from scratch and tossing out their client/server and mainframe apps, but rather are integrating these apps interfaced to new web applications. In most cases, the backend application is unchanged except for the new API and is likely to remain unchanged for some time since it still has a very, very large batch and/or internal function to perform. This doesn't exclude a tornado for products to support the web component or application and it suggests a likely tornado in products with which to do that integration, but we are a long, long way from a mass swap out.
As a first-mover
It would be difficult to class BEA as a first mover since Forte has been delivering tools for this class of application since before BEA was formed and their tools have been consistently and significantly superior to those from BEA from BEA's inception to the present. However, Forte's primary tool set is based on C++ deployment, not Java, which is the "hot" area, at least in the trade press. I make the latter qualification because there is some evidence to suggest that the number of applications actually being deployed with Java is much less than what one would expect from the "noise". Particularly with respect to the enterprise class, highly scaleable applications which would seem to be BEA's market, this is not especially surprising since Java execution performance is still at a 2-3X disadvantage to C++, to large a disadvantage to be acceptable in performance sensitive environments.
in the Java camp, which implied an allegiance to "openness"
This is an extremely importance aspect of this market. The slowness of the wide adoption of the demonstrably superior Forte C++ tools is blamed in large part on the "proprietary" stigma, despite the fact that Forte has long supported virtually every relevant interface standard including CORBA and Tuxedo. Prior to J2EE, proprietary extensions were virtually necessary in a Java appserver in order to get meaningful work done. With J2EE we still don't have standards that cover all the aspects which Forte covers, e.g., distributed events, but at least it is far enough along that one can make some pretense of doing enterprise applications within the standard. This has turned the enterprise Java market into a heavily standards focus competition -- components like appservers must be standard in their behavior and thus are limited in added value to areas like ease of use, scalability, analysis tools, etc. Similarly, EJB components must be standard and must run on anyone's appserver to be acceptable in the market. What is not standardized, however, are the development tools by which EJBs and the like are created.
It is very difficult to make any serious judgments about tornado or gorilla status when the figures are so radically conflicting. These are clearly apple to orange comparisons. You mention some likely sources of the differences such as whether or not services are included. One suspects that another major source is what products are and are not included in the revenue. For example, do BEA's figures include all of the sales of Tuxedo in the same category as Weblogic sales even though many of the historic uses of Tuxedo have not involved appservers or Java? And, if we are including Tuxedo, then are we including competing technologies for other companies that address the enterprise application integration market, even though that also very often has nothing to do with appservers and Java? Are we including application components like EJBs in the total? If so, are we including the revenues from all of the companies that make EJBs, but don't make appservers in determining market size? Are we also including the many products like those in the very broad iPlanet suite which are far beyond the level of individual EJBs, providing integrated sets of services to cover a broad range of e-commerce application requirements? Particularly since one can rarely, if ever, get a real product-by-product revenue breakdown from any of these companies, it is very easy to see how one could prove just about anyone was a leader depending on what was and was not included and what guesses one made about the breakdown. In short, if "Coleman claims that BEA has the 70% to 80% market share", my question is "market share of what?".
Moreover, I believe by the criterion of acceptance by a pragmatist majority
Adoption <> acceptance <> succesful deployment. It is one thing for a company to buy a limited user license for development and quite another to buy multi-thousand user licenses for deployment, especially in a standards-focused market like Java appservers where there is now a presumption that one can develop on appserver A and deploy on appserver B. Most adoptions of all of these products at this point are for development and limited deployment. Customer lists are notoriously bad indicators of what companies are actually using and using in volume. |
