Data Windows

In RioDB, a Window is a finite segment of data records passing through the stream. So if the stream is an unbounded (endless) series of data records, a window is a bounded segment of that series.

Examples of windows:

• Given a stream of stock trades, a window could be all bids on the stock TSLA during the past 10 minutes.
• Given a stream of requests sent to a web server, a window could be all POST requests received by a server during the past hour.
• Given a stream of GPS location coordinates from smart devices, a window could be the last 100 GPS coordinates of each device.
• Given a stream of sensor data from a motor, a window could be the motor’s temperature readings during the past one minute.

RioDB supports SQL-like commands to create and manage Windows.

If you are familiar with the difference between Sliding Windows vs Tumbling Windows, RioDB supports only sliding windows (because tumbling windows suck).

The primary reason for the RioDB project is to be able to run Sliding-Window computation in demanding situations where the data stream moves extremely fast, and where a Tumbling-Window’s lack of precision is unacceptable.