Time Series Forecasting with Telegraf

If you’re familiar with Telegraf, you know that you can easily configure this lightweight collection agent with a single TOML configuration file to gather metrics from over 180 inputs and write data to a wide variety of different outputs and/or platforms. You might also know that Telegraf can act as a processor, aggregator, parser, and serializer. For example, you might even be familiar with the Starlark Processor Plugin that gives you the ability to perform various mathematical operations in Telegraf.

Needless to say, Telegraf is pretty awesome. But did you know that you can also use Telegraf to perform forecasting on your time series data? The Telegraf Execd Processor Plugin is a processor plugin that makes Telegraf extensible in any language, allowing you to process your data however you need. In this blog post, we’ll learn how to use this plugin to continuously generate forecasts and perform anomaly detection. The repo that accompanies this tutorial can be found here. This blog is a recap of Anais’ (me) and Steven’s (a senior software engineer on the Telegraf team) November 2020 InfluxDays talk.

About the Telegraf Execd Processor Plugin

The Execd processor plugin runs an external program as a separate process, pipes metrics into the process’s STDIN, and reads processed metrics from its STDOUT. The…