Esper is a 100% Java component for CEP and ESP applications. Esper enables rapid development of applications that process large volumes of incoming messages or events. Esper filters and analyzes events in various ways, and responds to conditions of interest in real-time.

Esper works like a database system turned upside-down: first one supplies the queries, and then runs the data through. Esper tells you query results right away on a continuous basis, as soon as any data (events) or combination matches one of the queries and produces new results. So, for example, take q query like:

select zone, count(*) from RFIDEvent.win:time(1 min) group by zone;

The sample query above looks at a stream of radio frequency tag reading events and counts the number of RFID readings per zone for the last 1 minute of events. This can be useful in the RFID domain to monitor zones, for example. Or take one more example which picks more individual events out of an event 'cloud' via a pattern:

every first=RFIDEvent(zone = 1) \-> RFIDEvent(assetId=first.assetId and zone = 3) where timer:within(1 hour)

Now the "every" is an operator that indicates that a pattern looks at each RFID event and doesn't stop. The "->" indicates a followed-by relationship between events. The "where" is a condition that turns false after 1 hour. In total, the pattern looks at all RFID events that indicate that the RFID reading is in zone 1, and are followed by an event for the same asset id and now in zone 3, skipping zone 2! It's alerting of that condition as soon as it happens.

Interested in more? May we suggest the O'Reilly article on Esper: Event Stream Processing and Correlation