ESB: time to grow up

While some of its rivals have descended into a puerile debate about who was first, Sonic Software is showing admirable maturity in its handling of ESB evolution. But will it follow through to the logical conclusion and leave the acronym behind?

When people start attacking you, it's usually a sign that you're on a winning streak, so I don't suppose Sonic was too disappointed when Tibco Software's founder and CEO Vivek Ranadive launched into a tirade about Sonic's claims to have invented the ESB:

In an interview with ComputerWire, Tibco's Ranadive said: "That's a complete joke. I personally invented it. We've always called ourselves 'The Information Bus Company', and the ESB is no different ... Sonic came in and made fictitious claims. Having one server and two clients and claiming you can do an ESB — they don't come near it. You need to have a whole integration stack ... You can only obfuscate the truth for so long."

Unfortunately for Ranadive, Gartner's resident ESB expert — and the one person, who if anyone, was first to independently validate the term — was quick to put the record straight, as Jason Stamper, editor of ComputerWire's sister title Computer Business Review, revealed in his blog:

"What Tibco was selling in the 1990s was an enterprise message bus, not an enterprise service bus," argued Schulte. "Like [Candle Corp's] Roma [a rival ESB ancestor], it predated web services so it was not an ESB ... Tibco’s products in the 1990s were supersets of the first ESBs in many respects but their lack of web services support meant that Tibco had a subset of an ESB in some respects. Now, of course, Tibco’s Businessworks product family has web services support so it is a pure superset of an ESB."

Of course, Schulte is careful to couch his explanation in terms that provide a few crumbs of comfort for all involved, whatever their position. He does work for Gartner, after all. But he quite clearly contradicts Ranadive's core assertion to have "personally invented" the ESB. Now if Tibco's founder and CEO can't (or won't) recognize the difference between a messaging bus and a services bus, what does that tell us about the clarity of the company's strategic direction? No wonder Ranadive is fuming.

And yet the timing of his attack on Sonic is in fact probably an even worse lapse of judgement than the attack itself. If Tibco is sweating over upstart ESB competition, then it ought to worry even more about a whole host of new arrivals that are undercutting Sonic by orders of magnitude, let alone Tibco. Loosely Coupled has just published the latest edition of our paid newsletter, which contains a couple of articles that surface some very interesting developments pertaining to ESB.

The first of these is the formal approval of Java Business Integration 1.0 at the end of July. There's been quite a lot of confusion about JBI, which is odd really, because in actual fact it's quite simple. It's a SOAP-based ESB for the Java platform. There are already three open-source implementations under development; Iona's Celtix, the Apache-licensed ServiceMix, and Sun's own OpenESB. Why pay Sonic (let alone Tibco) for ESB capability when you can get it from a variety of reputable sources for free?

The second is the whole XML-aware networking trend that Cisco's AON announcement has lit a fuse under. Sonic's own Hub Vandervoort was surprisingly welcoming of the notion that the likes of Cisco might put core ESB functionality on a chip and embed it in a network router. A whole load of other functions would still have to be provided by the ESB software, he explained, such as orchestration, monitoring and metadata. So Sonic would still have a ready market for its products, even if some of the basic messaging work was delegated down to network devices.

What I think all this reveals is that sooner or later Sonic is going to have to fess up to is that what it's aiming to offer is not really an ESB so much as a complete packaged SOA. And that's going to provide a problem of perception. Because if Sonic keeps on marketing its product suite as an ESB, at the same time as a whole load of other people put the ESB label on open-source infrastructure offerings that do a lot less — because they focus on the core message routing, mediation and persistence functions of an ESB — then people are going to start wondering why Sonic's ESB is so much more complex and costly (goodness knows, by the way, what they'll say about Tibco's).

But I think Sonic is being very mature about this — why else would it already have named its flagship bundle the "Sonic SOA Suite"? Sonic knows that it has to grow up into a fully fledged SOA infrastructure vendor. But it also sees the wisdom of making the most of its well-earned stature as a recognized leader of today's ESB market. One day it will indeed quietly move on from the ESB moniker. In the meantime, I'm sure it's making the most of all the attention it's getting from Tibco.

posted by Phil Wainewright 8:32 PM (GMT) | comments | link