neutrino 0.2

High Performance Redis Timeseries

Neutrino

A high performance library for storing scalar timeseries data in redis.

Installation

To install, simply type:

$ sudo pip install neutrino

Design

Time series are stored as scalar values in bins with a fixed time step. These bins are stored in chunks, and each chunk is stored under a single redis key.

Instead of specifying these parameters for each time series individually, the parameters are specified for a pattern, which is a regular expression. When time series data is later added, the name of each time series is matched to these patterns, and for each match a time series is added.

For example, the rule

>>> ts.add_rule(r'spam.*','f',1,10,100,'avg')

will match any key starting with ‘spam’. Any key matching this pattern will be stored in a time series which will average multiple values in a single bin, with a time step of 1 second, a chunk size of 10, and chunks will expire after 100 seconds.