Death to Latency: Building Reactive, Cloud Native Apps with Akka.NET
summary
Real-time cloud apps use in-process state and the actor model to beat database roundtrips: Akka.NET actors collocate data and behavior, persist deltas or snapshots to a DB, and use cluster sharding + Akka.NET cluster to guarantee a single owner per entity and transparent rehoming during failures. Deployments use rolling add/remove on Kubernetes to avoid split-brain. Reported throughput ranges from ~6M msgs/s on older hardware to 80–90M msgs/s on .NET 6 across nodes, and external clients connect via gRPC or WebSockets for low-latency delivery.
AI-written summary · about
chapters (10)
- 0:00Start
- 1:31Real-Time Web
- 5:51Real-Time Domains
- 13:29Why Old Paradigms Fail
- 16:53Low Latency Needs State Locality
- 19:20Application as Source of Truth
- 21:42What is an Actor?
- 33:12Key Business Questions
- 34:03Installing Akka.NET
- 54:34Deployments and Infrastructure
related talks
44mDeath to Latency: Building Reactive, Cloud Native Apps with Akka NET
37mAkka Present and Future
46mAkka Streams & Reactive Streams in action
40mTop Mistakes When Writing Reactive Applications
01h 11mReactive Architecture Patterns Debate
52mBuilding Reactive Apps
38mReliable Messaging in the World of Actors
40mReactive programming: lessons learned