To liven up the International Summer School last week, the organisers arranged for a veritable bullfight between protagonists for the Stream Processing, Rule-based CEP and Processing-by-Subscription worlds, represented by Pascal Felber, TIBCO, and co-host Hans-Arno Jacobsen, moderated by Budra Gedik.
Naturally I took the Switzerland route to neutrality and duly claimed that all the other approaches were in fact subtypes of rule-based processing. Although I could have tried educating the audience (and panel) on the advantages of Rete-type inference rules in a CEP context, it made more sense to claim the moral high ground and subsume the competition. On the other hand, I’m not sure my panel colleagues totally disagreed with my hypothesis, which made for a somewhat tame discussion afterwards.
Despite Budra’s best efforts, the audience could not be goaded into asking many questions, so we defaulted to his list including:
- Was there scope for new Event Processing Languages (EPLs)?
Answer: Probably, given that there are multiple language patterns today, there could be more tomorrow. In fact… - Was there a clear winner between declarative and imperative approaches to EPLs?
Answer: No: this is a compromise between maintainability and ease-of-use, and both are needed… - What about static versus dynamic typing in EPLs?
Answer: Mostly we trend to static for performance reasons, although dynamic might be better for rapid prototyping. - What about the need for modularity?
Answer: Yes, in order to drive manageability for re-use, collaboration, verifiability, re-factoring etc. - What can be done to solve interoperability?
Answer: It’s already solved… by passing events between systems!
Slides presented in this panel can be downloaded from here.
