Skip to content
memoDwarves Foundation

Research notes/Languages

Scenarios using concurrency in elixir

2 min read

Fire and forget

Tell the system to execute some code asynchronously and not care about when it finishes, nor the result.

Solution: Task abstraction in elixir. Tasks are processes meant to run a single action within their life-cycle

Practice:

Run a one-off action in Elixir to spawn the tasks under a supervision tree. We can easier to manage the processes(clean)

Task.Supervisor.start_child/2

The default strategy for the Task.Supervisor is :temporary - never restarted.

Implementation:

  • In the Application we add a Task.Supervisor, is named such as: AwesomeTaskSupervisor
  • When we spawn a process, we start a child task with AwesomeTaskSupervisor name.
Task.Supervisor.start_child(AwesomeTaskSupervisor, process_func)

Fan-in/fan-out

Tell the system to execute consists of multiple operation. We run asynchronously, but care about the results. Example: uploading a bunch of docs to S3, sending a batch of emails to people.

Solution: Task without using a supervisor

Implementation:

  • Using Enum.map/2 and Task.async/1 to generate the processes.
  • Using Enum.map/2 and Task.await/1 to wait all of processes are done

Scheduling work

Tell the system run some work periodically every N minutes/hours/etc.

Solution: GenServer abstraction. We save the period rule in the genserver. Using Process.send_after/3 to trigger the handle_info/2 processing function.

Library: Quantum

Caching data

Tell the system cache the frequently data. Such as: caching the frequently response, temperature data.

Solution:

  • Agents abstraction: a single process, therefore, we got bottleneck
  • ETS: faster, support concurrency reads and writes
  • GenServers

Library: https://github.com/sasa1977/con_cache

Reference