Astarte Flow base computation unit is a block and their purpose is consuming, producing and processing messages.

Blocks are used as pipelines building blocks by chaining them together, in order to define a logical computation topology.

There are 3 kinds of blocks that play different roles in a pipeline:

  • Producer (Source): they produce messages from events coming from the outside world
  • Consumer (Sink): they output data from Astarte Flow to the outside world
  • Producer & Consumer (Transform / Process): they transform messages

Blocks Configuration

Blocks behavior depends on their configuration, that can be changed by setting some properties ( e.g. to_json block has a pretty boolean property which allows deciding wether JSON output is human readable (pretty) or not).



Blocks configuration is validated against its JSON Schema, therefore a new block implementation requires writing a JSON Schema for it.

Implementing a Block

Blocks can be implemented using different technologies:

  • As a Docker container using any language suitable for Docker (e.g. Python), which is the best option for implementing custom complex algorithms
  • As a chain of existing blocks (a pipeline), that is the go-to solution when a block can be implemented by just chaining blocks together
  • As a Lua 5.2 script, that is the idea solution for small adjustments and simple data transformations, such as applying a conversion formula
  • As an Elixir module, that is how built-in blocks are implemented