Came back today from a short vacation in the Grand Vista Boutique hotel in Kfar Yuval, at (almost) the most northern (and east) part of Israel. Looking at some related blogs, I found Joe Mckendrick's blog mirroring Colin Clark saying that the "CEP vendors" failed to produce the multi billion dollars revenues they promised to their investors this is due to two facts: 1). end users cannot understand it; 2). it does not provide the IT people a tool to build a complete application.
I am sure that Colin, who has been a key person in one of the first generation startups in this area is right in his perspective. However, coming from the big blue, I have a slightly different perspective.
I'll start from the last point, since this is a key to the rest: the claim is that the "CEP vendors" did not offer a tool to build a complete application. This is of course true; there are applications (indeed in capital markets but also in other industries) that have applications that are mostly pure event processing applications, however, this is not the major part of the market; the major part of the applications market is such that require event processing capabilities within an enterprise computing platform, that also includes BPM, BRMS, ESB and other stuff. This is the approach taken by the bigger companies that got into the event processing game. It is true that a start-up cannot do it alone, and start-ups that remain "pure event processing" companies are confined to the niches of stand-alone EP applications (in which the potential was also not reached, so there is certainly a place for products in this niche). As for the multi billion dollars, I don't think that there is any analyst who predicted multi billion dollars for that market in 2010, it takes time to develop a market, and as a stand alone EP products, may take longer time to get there. When we did the business evaluation of the EP market around 3 years ago, we got to the conclusion that for the market of "EP capabilities within an enterprise computing platform", one can look at accounting value or on economic value and they would yield different results. Let me give a simple example in order to show what I am talking about: let's say that a big enterprise is doing a software deal with the value of $15M for its enterprise computing infrastructure. This may include many software components licenses. Let say that the deal is a bundled deal, but if we'll try to isolate the value of the EP capabilities license, we'll get to $500K. However, in the RFP the event processing capabilities are required part, and a vendor that does not have the required capabilities is disqualified for this bid, then this vendor, losing the bid did not loose a $500K deal, but lost $15M deal. This is the distinction between accounting value and economic value. We see more and more RFP that either put event processing capabilities as mandatory, or give some weight to it as explicit criterion, this is done, even if the intended use of EP is planned in 2-3 years in the enterprise IT investment plan. I think that looking at economic value (e.g. deals that have EP components inside) we might already passed the 1B dollars mark, but this arithmetic, of course, makes sense to vendors who own enterprise computing platforms and not to start-ups that has to earn the actual accounting value of stand-alone EP applications (or sell it as embedded solution). About the "not appealing to end users" - I'll write a separate posting about it; some of the analysis from the user's point of view was done as part of chapter 1 of the "event processing manifesto" (a follow-up to the Dagstuhl seminar on event processing) which is planned to be published in January 2011, and presented in March 2011 within the virtual EPTS symposium. More about this topic - later.