Erlang & Akka Actors - A story of tradeoffs
summary
Compares Erlang/BEAM and Akka actors on the JVM, examining how their runtime and implementation choices trade off performance and fairness for the actor abstraction. Covers concrete factors such as BEAM’s per-process model and garbage collection design versus Akka on the JVM, the impact of type systems on efficiency, and how preemptive scheduling influences long-tail latency.
AI-written summary · about
chapters (19)
- 0:00Intro
- 1:27Structure
- 8:10Actor Model - Intuition Dining Philosophers
- 11:38Messages: Send by Copy
- 18:56Configurations
- 23:10Elixir philosopher
- 24:53Implications
- 25:37Scheduling in Akka
- 27:40Dispatcher: How does it work?
- 29:18Let's go back to the Akka Philosopher
- 29:58Throw more threads?
- 31:27Solution 1
- 32:59Solution 2
- 36:09If the threadpool was just
- 37:53Dispatchers are configurable
- 40:02Futures gotchas
- 41:56What's different?
- 46:45Akka Behaviors
- 47:55Observation
related talks
25mSCALAZ 8 VS AKKA ACTORS
45mSo you want to be an actor? Lessons from implementing the actor model
40mComparison and Commentary on "Application Operating System"
44m10 Lessons From Implementing The Actor Model
51mThe Erlang VM or How I Started Loving Parallel Progamming
22mActing Lessons for Scala Engineers with Akka and ZIO
54mScala concurrency patterns with Actors
43mConcurrency Oriented Programming in a Modern World