10 Lessons From Implementing The Actor Model
summary
Production system uses the actor model with Akka/Akka.NET in a cluster, replaces per-file persistence with media set aggregate persistence (TV/radio program as the event-sourcing root), and removes a file dispatcher by assigning one actor per file worker. Observed performance: 4 million msgs/s on tell vs ~150k with ask, so avoid ask and prefer one‑way messages plus FSMs for aggregation. Other concrete choices: start non-clustered, keep persistence at aggregate roots, let actors schedule reminders and validate themselves, avoid singletons and be conservative about split‑brain resolvers.
AI-written summary · about
chapters (16)
- 0:00Intro
- 4:51Why actor model & Akka?
- 8:39Lesson 1
- 15:39Lesson applied
- 16:54Example
- 19:42Lesson 2
- 23:09Lesson 3
- 25:05Lesson 4
- 27:55Lesson 5
- 29:44Lesson 6
- 34:16Lesson 7
- 35:56Lesson 8
- 38:19Lesson 9
- 39:53Lesson 10
- 43:42The 10 lessons
- 44:06Outro
related talks
45mSo you want to be an actor? Lessons from implementing the actor model
49mDDD, Event Sourcing and Actors
54mScala concurrency patterns with Actors
59mActors in clusters: why, when and how
27mFour Principles of Reactive Architectures
54mIntroduction to the Actor Model
52mBuilding Concurrent, Fault Tolerant, Scalable Applications in F# using Akka Net
01h 03mErlang & Akka Actors - A story of tradeoffs