Having read through the Fool post responding to your extract, it seems to me that much of what he has to say boils down to the fact that his focus is the e-business platform market, not the appserver. This has lead to a certain amount of seeing difference where none exists since a number of my comments were in response to a contention that the appserver was the thing, not my trying to claim that the appserver was the thing. One of the more conspicuous of these is blaming me for the category definition problem. But, such is the way of things when one has these multiple threads cross connecting, I suppose.
A couple of specific points:
the category that I believe is now in its third quarter of a tornado is the eBusiness platform.
Based on three quarters of 100% growth? At what revenue level? Given how ultra new most of these products are, I am inclined to some caution before announcing a tornado since, after all, it is very easy to double revenues when one hasn't sold all that much in the prior year or quarter.
Some of these issues are very tricky and when solved well provide a significant barrier to entry. BEA, for example, does a good job of clustering upto 8 servers together.
I can't speak to relative capability since this is not an aspect I have looked into, but I do know that the iPlanet appserver supported clustering over a year ago. I think this quote from a ZDnet review is apt:
With six-month release cycles, the app server product with the edge is usually the one that was released most recently -- the J2EE standards push is leveling the playing field, and, frankly, we see more similarities than differences among Java-focused products on the market.
Despite his characterization of Clustering on this scale is rocket science I guess I will remain a little skeptical that BEA has engineers who know about scaling in a way that no one at IBM or Sun do, considering the extremely high end servers that both companies produce.
One reason why and ISV, VAR or customer wants to standardize their eBusiness platform is so they need only implement this layer once ... The developer needs to change the underlying hardware and systems environment and have this change be transparent to the applications.
This got a bit fuzzy for me since it is still in the section on the "below the appserver" part, which seems to me to be simply "how well does the appserver actually work", not a part of any e-business platform, unless the appserver itself is considered a part of the platform and I thought we were trying to keep these separate. I would agree that people were attracted to the idea that the e-business part could be implemented without concern for hardware and OS details since this part was handled by the appserver. Moreover, in the Java market specifically, I think people are looking increasingly to being able to move applications across appservers since they may be selling into environments with existing appservers.
This leads to a discussion of the second layer of the eBusiness platform, the business application components and the eBusiness applications. .... This list is very, very long. This is another area where BEA has made significant progress (as has IBM).
Among others ... as I said, I find the iPlanet offering extremely rich.
The WebLogic Portal Server, Personalization Server, Commerce Server, Campaign Manager are all proprietary components built up from the standard J2EE framework. IBM, Oracle, iPlanet, Bluestone all has their own versions and nobodies' components are compatible with anybody else's components.
No disagreement. Without having done any detailed comparisons, I am willing to be that many of these components or subsystems are also reasonably similar in capabilities, particularly if one allows to the latest release rule cited above.
However, this does not mean that there are not some components or subsystems which are highly different in capability. The question is which ones and how strategic are they. If two vendors both have a basket full of the basic e-commerce tools, chances are that one will not be that compellingly different than the other, not in a gorilla type of compelling. In fact, given the necessary similarities of things like shopping cart components, it might not even be totally horrid to have to switch at some point. The question is whether there are components which are not shared or which are shared but substantially different in capabilities, components which one would have to do without if one switched or failed to make that choice.
Again, I haven't had occassion to do a point by point comparision, but one possible candidate for such a differentiator is the iPlanet Integration Server, formerely known as Forté Fusion. I believe this product is significantly ahead of even the best that come from other dedicated EAI vendors, much less those offered by other appserver vendors. The question is, how important is this. iIS has certainly become a flagship product for iPlanet and I think for a very good reason. Development of new Java e-commerce applications is only rarely happening from the ground up without a surrounding context of other applications in varying technologies. E.g., an insurance company may want to starting selling insurance on the web, but they already have all of the risk assessment software, policy management software, billing software, customer management software, etc., none of which have anything in particular to do with the web, but all of which must interface with that web application. This is where an EAI tool comes into play and the capability of that tool can make a major impact on the overall cost to implement.
There are other, secondary, proprietary layers - PDA support, Wireless support, development tools, UML modeling, business intelligence reports, etc.
Which may or may not all come from the same vendor. An increasing number of companies, for example, are using Actuate for the web-enabled reporting aspect.
One area where BEA and IBM have clear leads and high barriers is in transaction processing. BEA with Tuxedo and IBM with some version of CICS are undeniably leaders in providing mission-critical transaction services.
Agreed that one has to support at least one and preferrably several distributed transaction systems, but I don't know that one has to be the primary vendor for that system, particularly since it may already be in place in the kind of scenario I described above.
The key value chain players have voted overwhelmingly for BEA.
They have a nice list ... does it represent a substantial number of all of the potential key value chain players?
If you can look at the BEA value chain and still argue that they do not LEAD in the solution market I would be interested in seeing that argument. It is the solution presented by the value-chain that customers buy. Remember, the BEA value-chain push is just six months old.
I would suggest that something that is only six months old is a little early to call.
How do you explain the rate of license growth for WebLogic? These systems are being actively deployed today. It is just the beginning but deployment is happening.
It is the fact that it is just the beginning that I find significant. How many systems are deployed with over 1,000 users? Over 10,000? Of the total number of current sites, how many have fewer than 100 users? What percentage of production sites are ramped to full planned capacity? What percentage of the total base of this type of system which will exist in 5-10 years is deployed and in production today?
You cannot say that J2EE is in charge of the architecture. You can say that WebLogic controls the architecture. Once again this is the appserver vs. eBusiness platform issue.
I agree that this distinction is very important in this discussion, but where is the evidence that the WebLogic e-business platform architecture has established itself as a de-facto standard. Do IBM or Sun or anyone else make compatible add-on or replacement components or do they have their own architectures? Popular? Yes. A defacto industry standard? No. |
