The Dapr Actors Journey: From Understanding to Intuition
summary
Dapr implements the virtual actor model where actor identity is type+ID, state persists in external stores, and actors execute single-threaded with isolation; actors support reminders and timers and scale into millions of actors. Placement is handled by a placement service with Raft (HashiCorp): three servers elect a leader, sidecars open bi‑directional gRPC streams, leader sends lock → table → unlock to guarantee deterministic, consistent placements. SDKs exist for C#, Java, JavaScript, Python and Go; common uses include IoT devices, game sessions, and NASA glove inspection.
AI-written summary · about
chapters (12)
- 0:00Introduction to Dapr actors
- 0:39Defining the actor model
- 3:03Actor model history
- 4:00Key actor properties
- 5:49Virtual actor model explained
- 10:26Dapr implementation
- 12:20Placement service mechanics
- 17:24Placement service architecture
- 19:54Practical actor use case
- 23:31Interacting with actors
- 24:32Advanced use cases
- 30:53Final thoughts
related talks
49mDapr in Practice
30mDapr case study: Man Group
04mProject Lightning Talk: Dapr: Start Building Distributed Applications with Ease Using Building Block APIs
31mBuilding scalable distributed systems powered by Dapr OSS.
37mDistributed Actors and Where They Live
48mAbstracting My Distributed Architecture? Why and How?
51mDay 2 Session 2
14mData-Parallel Actors: A Programming Model for Scalable Query Serving Systems