With a name like 'Liquid Computing', you'd think BEA would try to avoid accusations that its newly launched SOA strategy is just so much vaporware. But its composition still seems to be more promise than substance:
Project QuickSilver is a plan to develop an enterprise service bus (ESB) offering that will also boast SOA management capabilities. Adding SOA management on top of ESB is a sensible proposition, but since BEA currently offers neither technology it's quite an ambitious undertaking to develop from scratch. Perhaps this is a hint that BEA is going to be on the acquisition path for some ready-made XML transformation and services management technology.
Project Alchemy is a technology concept that BEA chief architect Adam Bosworth began outlining when he took up weblogging last year [link updated July 2004]. The aim is to create a browser-like user interface that lets the user carry on working when disconnected. The project is well named, for this quest truly is the philosopher's stone of Internet computing a platform that will turn low-cost devices into Windows-like clients. BEA is the latest in a long line of aspirants to embark on this quest. None have yet proven successful, so BEA is taking quite a risk in making it a cornerstone of the strategy.
The only substantive product is BEA WebLogic Server Process Edition, due to ship this summer, which will help developers assemble services from across an SOA to form composite applications. Even this comes with an undated promise attached, that later on it will add process assembly capabilities that business analysts will be able to use.
Although not part of the Liquid Computing line-up, it's also worth mentioning Project Beehive, announced last week, which is a commitment to place some of the technology surrounding the WebLogic Workshop development tool into open source licensing. BEA hopes this will bolster its standing with developers who have started to introduce open-source application servers into their infrastructure alongside BEA's own WebLogic platform (or in some cases in place of it). BEA wheeled out web guru and publisher Tim O'Reilly to present a keynote on how software companies can develop the equivalent of an "Intel Inside" strategy to win commercial advantage within the emerging open source architecture of service-oriented computing. Microsoft Indigo architect Jon Shewchuk was using the word "platformization" to describe a similar process in an interview with internetnews.com last week Microsoft of course being the past master of this art, the secrets of which BEA hopes that Bosworth and his cohorts have brought with them from their former employer.
InfoWorld's Eric Knorr was blunt in his assessment: "Liquid Computing is basically BEA's answer to IBM's On Demand initiative. In other words ... primarily a marketing concept." Although essentially correct, I think that's a trifle harsh. BEA needed to make a clear statement of direction on SOA. When Loosely Coupled reviewed BEA's SOA strategy last October (see BEA stakes future on SOA adoption) its position was basically, 'We think SOA is a good thing, so buy all the bits from us.' As a platformization strategy this was, to say the least, somewhat naive. Now at least customers know what bits they'll actually be getting if they align themselves with BEA. But there's still too much aspiration and too little deliverable substance in the strategy. When CEO Alfred Chuang tells InfoWorld that "We can't just say product, product, product anymore. We have to explain the whole picture," he's admitting a candid truth: BEA has to explain the whole picture because it can't yet offer any product, product, product.