vincent
09-20-2008 07:20 AM
EPTS4 day 2 started with a a keynote by Susan Urban of Texas Tech Uni on a Roadmap for Research Directions. Susan comes from the active database world and was talking about fun stuff like distributed event processing agents and so forth. Susan emphasized “intelligence” as the ultimate goal (and hence of course CEP’s use for “operational intelligence”). Susan mentioned for example her baggage being routed to arrive ahead of her as an example of this (which, coincidentally, is also a TIBCO CEP use case for one of our airline customers). Future research directions include things like formalizing the semantics of the event cloud, modelling validation and correctness, and future visualizations in dashboards.
Susan’s best quote for me was “you need an event processing language that embeds SQL, not the other way round”. This directly mirrors our thinking with
TIBCO BusinessEvents embedding
OQL-based continues queries for dynamic set-based facts.
There followed a CTO and Researchers panel:
- Jerry Baulier (Aleri) commented on trends such as
- the convergence of rule and “relational dataflow” paradigms
- multiple authoring levels (as for various levels of business and technical user)
- “hybrid analytics” which included enriching event processing with data (actually something that has always been a part of event processing)
- advanced visualization
- Mani Chandy of Cal Tech talked about the social aspects of event processing - such as personal event processing. This is clearly a untapped area (and came up as a question at workshop earlier this year). Probably the next ad-funded Facebook / Google phenomenon will be in this area - think Skype or BitTorrent for personal event mapping / tracking / decision support.
- Malcolm Lockhart (EventZero) commented on event processing domains being:
- future internet such as semantic web events
- “special” areas like military intelligence
- business, such as BPM and MDM
- industry, such as manufacturing processes
He also sees the topology of event processing networks growing from monolithic to distributed to event-cloud based, with increasing complexity and numbers of event sources and sinks.
- John Bates (Apama) extended Mani’s theme with event processing for augmented, and maybe even extented, life. Things like drive-by information provision, and pill-based body instrumentation. His vision includes events as first class objects (which is a goal of the forthcoming OMG EMP RFP, for example), realtime BI for the power user (marrying the power of something like Spotfire with CEP like BE), and federated components (such as distributed complex rules tracking millions of event objects).
- Albert Mavashev of Nastel talked about using CEP to monitor SOA, transaction monitoring, and CEP as a cloud service. The former are already reality, of course, and the latter is probably something that will be allied to SAAS and the social EP.
- John Morrell (Coral8) talked about the usual software feature metrics like power (eg via grid computing), control (eg via statistics), deployability (eg managing the event processing network), user empowerment (eg business user interfaces), and integrations (eg adapters).
Note that if TIBCO had been on the panel, we wouldn’t have presented anything vastly different - pushing the envelope on all the above. Maybe some more on future event distribution mechanisms and the importance of performance networks, data grids, the management of automated decisions, integration with processes, and maturity through standards.
So this was pretty much a concensus, with the vendors going for making event processing more slam-dunk solutions for business and IT needs, and Cal Tech going for the non-business next-Google model.
In the Q&A David Luckham asked whether the internet can exploit content-based routing? I would have answered that CISCO is already into rule-based event routing and rule engines, and appropriate
rule interchange standards are around the corner, so why not?
Source...
09-20-2008
