vincent
Tue, 05 Feb 2008 15:38:48 +0000
One of the problems with new technologies is the issues of “over-hyping”. This normally occurs when the level of excitement exceeds realism, and claims are exagerated by analysts, journalists, and profiteers. In the ’80s and ’90s a prime example of this was the hype around AI, alongside taxpayer-funded grandiose research plans such as Japan’s “Fifth Generation Computer Project” and the UK’s “Alvey Programme” (although the US decided instead to fund a “Star Wars” programme - the jury is still out on which nation got the best ROI). Although “expert systems” and “knowledge-based systems” advanced the state of the art, they did not lead to widespread machine intelligence (or arrest any increase in automated stupidity). Funding dried up, the IT world returned to SQL and C/C++, and commercial rule-based development systems continued to evolve, albeit at a less frenetic pace, whilst addressing more mundane “business rule” problems.
Fast forward 2 decades, and the CEP industry is now in a similar situation to the AI industry. Analysts are getting excited, Wall Street has adopted the idea [*1], signal processing software specialists are saying that they were right all along, and even senior business people can understand the benefits and rationale of real-time event processing outside of the restrictions of sequential BPEL-type thinking. Is CEP in for a “fall” like AI did?
The answer, I believe, is “no”. For starters, there are no huge government R&D programs or taxpayer bets being placed on the technology, fuelled by hype-filled justifications. Secondly, the technology is not promising something that it hasn’t already delivered. Thirdly, the technology and tools use adaptations of existing, tried and tested techniques [*2]. And are for the most part supported by existing standards. Fourthly, you don’t need a PhD [*3] to understand how CEP works, or how it applies to real-time business scenarios.
Indeed, CEP is still relatively unknown compared to BPM and SOA. But it stands well on its own 2 legs, without any help from the hype-ers.
Notes:
[1] Also the title of this BCS SPA event in London, which we hope to attend…
[2] Tools like TIBCO BusinessEvents use common standards - in our case a standard class model, state model, production rule engine, and middleware connections from the SOA world.
[3] Lucky for me - but no doubt there are research topics available for advancing areas in CEP…


Source...