Taming & Composing High Performance Stream-Oriented Processing Systems
summary
Builds real pipelines in C# and F# with Akka Streams and the Reactive Streams protocol and shows concrete numbers: a CSV stock-ticker replay spikes memory from ~800MB to 2GB, naive time-based throttling drops messages, and switching to demand-driven backpressure via request(n) makes producers match slow consumers. Implements broadcast/merge graphs, uses mapAsync-style parallelism (examples: 10 vs 1), replays 70MB of tweets, calls an external weather API and runs sentiment scoring, and publishes runnable code on GitHub.
AI-written summary · about
chapters (13)
- 0:00Start
- 4:16by reactive programming?
- 4:49Rise of Event Stream Systems
- 5:29at is a Stream?
- 15:19sh Model
- 17:09Actor is a single and independent unit of computation
- 20:26m Actors to Streams
- 21:45or pipeline of functions
- 24:21aph DSL in C#
- 24:59eam - Operator fusion
- 27:55namic Push-Pull
- 28:51tive Stream interfaces
- 30:31Pipelining
related talks
39mAdvanced Stream Processing on the Edge
24mSwimming in the stream...
41mAkka streams for high throughput data processing
43mConnecting Reactive Applications with Fast Data Using Reactive Streams
22mHigh throughput data processing with GenStage
28mSweet Streams Are Made Of These
42m"Distributed stream processing in Ruby"
16mEmpowering Fearless Data Engineering