#ORF09 What’s Different about Rules in CEP Presentation

This entry is part 17 of 30 in the series October Rules Fest 2009

Paul Vincent talked about Complex Event Processing (CEP).

What is Complex Event Processing? It is continuously processing events (asynchronous events) and passing it through to business events. One of the goals is to find patterns in the events from the “event cloud”.

The event cloud is the sum of all available events. There are also event streams. CEP uses pattern detections on the clouds and stream and their histories.

Problems solved: situational awareness, “sense and response” and “track and trace”.

From a technical point of view what does CEP need? Events are transient, so they have a limited time to live. Some are immutable, etc.

Event examples:

  • SOA Service Requests
  • Scans (RFID, etc.)
  • Web requests
  • Messages, Packets (networking world)
  • Data streams (financial world)

Pattern matching of events is required. This should obviously resonate with the business rules community with pattern matching algorithms.

There is also a need for persistence and failover.

CEP Rules requires:

  • Temporal awareness (General patterns, state patterns, event state patterns, object history patterns)
    • A Simple Rule Model (similar to the PRR model) (Rulesets, Rules, variables, conditions, actions, events)
    • Rule Types (Basic condition-action, Triggers Event-Condition-Action, Timers and schedulers, event lifecycle, state transition)
  • Low latency and scalability

CEP uses State Models and State machines to implement the event life cycle.

Interesting presentation on CEP. I guess I will have to look at what products like Tibco and Drools Fusion offer for event processing.

Tags: , ,

Leave a Reply