Understanding Streams and Relations

Streams and relations are key concepts in the Intelligent Event Processor (IEP).

Streams

A stream is a series of timestamped events that have the same schema.

Assume that a stream has the following characteristics:

• Each event represents a stock transaction.
• The schema consists of the stock symbol and the stock price.

The following table shows an example of the events in the stream during a brief interval of time.

Relations

A relation is a collection of events that match a user-defined condition at a point in time.

You can define the condition in various ways. For example:

• All events that arrived in the last five seconds
• All events that arrived in the last five hours
• The most recent two events

Assume that you define the condition as all events that arrived in the last five seconds. For the example in the "Streams" section, the relation at time 2008-12-15-T10:35:00:000-05.00 would consist of the following events.

Let's move forward one second in time. The relation at time 2008-12-15-T10:36:00:000-05.00 would consist of the following events. Compared with the previous relation, one event has dropped out and one event has been added.

The relation at time 2008-12-15-T10:37:00:000-05.00 would consist of the following events. This relation has the same events as the previous relation.

The relation at time 2008-12-15-T10:38:00:000-05.00 would consist of the following event. Compared with the previous relation, one event has dropped out.

The relation at time 2008-12-15-T10:39:00:000-05.00 would consist of the following events. Compared with the previous relation, one event has been added.

A relation can be empty. For the example in the "Streams" section, the relation at time 2008-12-15-T10:45:00:000-05.00 would not contain any events because none of the events arrived in the last five seconds.

This page (revision-16) was last changed on 08-Jan-09 10:59 AM, -0800 by Jeff Stein.
This page was created on 15-Dec-08 09:18 AM, -0800 by Jeff Stein.

More info...