The note from Sam is highly instructive, because it shows the incompleteness of most current SOA technology, methods and implementations. I repeat his comment below.
We must crawl before we can walk, however, the absence of a full conceptual foundation is not merely an oversight; it is an omission. The omission is due to the trend to dummy down all discussion about solutions to the data interoperability crisis. "The crowd hollers " we need software (to sale), not conceptual foundations!), and in doing so reveals the cause of the current wasteful expenditures at e-Gov and elsewhere.
A proper conceptual foundation must start with systems theory, and a form of systems theory that is capable of modeling real time event interactions. These events, in natural reality, are complex in the sense that is not captured by deterministic or reductionistic paradigms. The argument has been made by many scholars, non perhaps better than Sir Roger Penrose. The work by category theorist Robert Rosen is one area where deep work on anticipatory interactions and bio-formalisms occurs.
If one starts with a conceptual foundation in system theory there is a simplification of the resulting IT architectures due to actually aligning the conceptual foundation with how events are generated in the real world, and how interactions between events develop (are orchestrated by real events and the interactions).
"Event Driven Architecture" as produced by the IT sector is predictably shallow, in precisely the way artificial intelligence and OWL type ontological models are shallow.
I have looked at the Complex Event Processing Blog
and will try to participate in this blog.... as perhaps now (2007) a more complete conceptual foundation to SOA might be tolerated by the e-Gov and other centers of power in federal use of SOA.
A new school has been developing called "second school"
Any suggestions as to how to advance this new school are welcomed.
Dr Paul S Prueitt Associate Professor and Chair of Mathematics Talladega College
Subject: Re: [soa-forum] Fw: Proceedings of NCOIC Session on "Event Processing for Net-Centric Operations" Date: October 3, 2007 8:13:22 PM CDT
It seems to me that EDA is largely a "polling" or "monitoring" activity in the popularized manifestation of SOA (i.e. SOAP, WSDL/REST, UDDI). That is to say, the events are largely based on checking and re-checking static, location dependent services. It seems inefficient and not scalable; and, therefore, perhaps not prevalent.
On the other hand, SOA implementations that implement dynamic services managed by dynamic infrastructures enjoy the ability to also leverage [distributed] event models inherent to the service infrastructures. A consequent robust event model can greatly enhance our ability to *manage* the services themselves and to realize more scale, adaptability and agility - at runtime.
There's another realm of event processing that is bringing tremendous utility to business analysts. Some call it "Complex Event Processing (CEP) for Humans". CEP is being leveraged to correlate events across sources and time - even non-events. Applications of this sort monitor (and/or listen to) myriad sources ( e.g. sensors, databases, feeds, etc.)
The more dynamic the constituent parts of the architecture become, the more proactive and resilient the various functions become that interact with the parts. Additionally, EDA becomes more of a "listening" or "push" activity. This will help reduce "noise" and bandwidth consumption, and will enable more intelligent systems.
I guess the net of my thoughts is that EDA becomes more apparent, relevant and prominent as SOA implementations become more dynamic.
Thank you! Sam
|