Let's replicate state together
summary
Claims a usable replication stack called replikativ built over three years that runs on JVM and browsers using Clojure/ClojureScript and core.async, CRDT-based eventual consistency, and simple pub/sub over websockets. A trivial chat prototype takes two to three hours to assemble; a Twitter collector produces ~100,000 commits and ~1,000,000 tweets for local analysis. Backends include memory for dev and Redis for performance; conflict-free convergence and offline writes demonstrably work.
AI-written summary · about
chapters (21)
- 0:00Intro
- 0:53About me
- 2:03Overview
- 2:52Engineering
- 4:37Distributed systems
- 5:54Linear order
- 9:03Prototyping
- 10:53Example application
- 11:42How is this done
- 12:35Closure script
- 15:00Write operation
- 16:15Build time
- 16:29Architecture
- 17:16Data types
- 18:58Implementation
- 21:03Crossplatform
- 21:26Abstractions
- 24:08Safety Performance
- 24:24Vision
- 24:55Business model
- 26:25Clone application
related talks
48mThe simplest way to build resilient applications
29mAbout Combining Functional Reactive Programming and Replicated Data Types
45mThe simplest way to build resilient applications
22mTear down this wall – rethink distributed systems
57mReliability Nirvana
20mExploiting Commutativity For Practical Fast Replication
28mNew insights into distributed and concurrent programming
25mCloudstate—Towards Stateful Serverless