Interesting to see that James Taylor is planning a Decision Management platform report this year (2012). James is particularly focused on the relationship between analytics and rule / decision execution technologies, which are of course increasingly popular business optimization techniques [*1]. The report prospectus defines 3 main technology areas of interest: rule and decision engine capabilities (c.f. TIBCO BusinessEvents and ActiveMatrix Decisions), predictive analytics or data statistics capabilities (c.f. TIBCO Spotfire, Spotfire S+, and Spotfire Miner), and optimization technologies.

In the latter category of optimization tools there is an interesting extension to S+ available called S+NuOpt. This provides techniques like LP and MIP, QP, and MOP to allow solving of constraint-rich problems like planning and routing. Traditionally these have been complex, long-running, CPU-intensive algorithms solving particular problems like “what airline routes should we offer this Summer”. But there is increasing interest in applying these resource (asset, personnel etc) optimizations against more and more exception events (such as the closure of some airports due to weather) - termed event-driven optimizations - and where the world of near-real-time event processing and operational decisions push up against the long-term strategic decision processing. Having cloud-based on-demand computing resources for irregular (but still computationally expensive) optimization computations, invoked as required through the assessment of complex events, is likely to be a key capability in future.

Although TIBCO Spotfire is commonly used to report on, and analyse results from, event processing using TIBCO BusinessEvents, I have not heard of any public case studies of NuOpt being invoked from CEP and event processing in an automated fashion. But if any usecases get published, I’ll add them here!

* Notes:

[1] See TIBCO Software 2010-2011 FY reporting on the success of CEP and analytics under “business optimisation”.

Although several folks in TIBCO had used the term “visual analytics” to try and describe what TIBCO Spotfire does, as opposed to merely reporting on views of data, the analysts have continued to position Spotfire as a “BI” tool. But now, the Spotfire marketing folk are starting to use the term “Spotfire Analytics” to better describe the positioning of the product, especially with the combination of S+ and Miner predictive analytics capabilities.

How does this affect TIBCO Complex Event Processing users? Not one jot, of course. Spotfire remains the preferred analytic tool for visualising the event data in a TIBCO BusinessEvents backing store (a.k.a. the “event warehouse”) and comparing that data with other sources. Using Spotfire you can identify new patterns that can then be used to update rules, queries and decisions in a TIBCO BusinessEvents application - or indeed, execute an S+ or Miner analytics model against the incoming events in either a regular, or special purpose, event processing agent.

Forrester’s James Kobelius made some interesting comments on the proposed buy-out of IDS Scheer - known for their high level process modeling tool Aris - by SoftwareAG. Such as:

“What’s most exciting, and potentially differentiating, about the Software AG/IDS Scheer BI portfolio is the combination of CEP with mashup and an in-memory architecture to support truly real-time, interactive analytics. In other words, Software AG/IDS Scheer could take a page out of the book of another SOA full-stack vendor: TIBCO and its Spotfire product group. “

Unfortunately the IDS Scheer BAM solution (which does not claim to be a Complex Event Processing solution) is OEMed - making full integration and exploitation more difficult. Otherwise, we agree, this could have had some potential to catch up with TIBCO.

Possibly more interesting to the event processing community is IDS’ alternative to value chain models - their event-driven process chain. Mostly this has been replaced by BPMN diagramming in modern process modeling environments, but nonetheless could still provide a useful high-level notation… perhaps SoftwareAG will propose it as a standard somewhere?