At SOAWorld in San Jose last week, I attended a session by Michael Rowley of Active Endpoints, covering Complex Event Processing (CEP), another TLA (three letter acronym) that is getting some press these days.  This was my first real exposure to CEP, so I’d like to briefly share my thoughts on it for others not familiar with it.  CEP is a smart way to associate events in the enterprise.  A typical scenario is credit card processing fraud detection.  While listening on an event stream of credit card authorizations, you might want to know if the same card was being used at 2 different physical locations at roughly the same time.  That’s a good indicator of fraudulent use.  CEP generally provides a query language, similar to SQL, but extended to deal with events.  The primary new dimension queriable with CEP is that of time.  For our fraud detection case, we might query for any credit card authorization occurring within a 15 minute window where the distance between physical locations is greater than 15 miles.  An engine processing such a query would listen to the event stream, and send a notification each time a match is found.  Thus, alarms are generated for potential fraudulent use.

CEP is a natural fit with business process management, like Active Endpoint’s line of BPEL engines.  For those that don’t know, Active Endpoints is a partner of XAware, as data services created in XAware fit nicely as the data layer in a BPEL process.  With Active Endpoint’s integration with CEP, events are automatically generated for each step in a BPEL process.  Then, the CEP query language enables operators, analysts, and developers a means to easily and flexibly evaluate the event stream, identifying important trends or critical compositions of related events.  This really seems to be a neat technology that I think will gain traction in the future.