Distributed Systems Wizardry
summary
Claims Golem is an open source serverless platform built on the web assembly component model that makes highly reliable distributed workflows 10x–100x simpler by providing reliable workers and reliable communication. It records external IO via a virtual host into an oplog and replays it to restore a worker to the exact instruction pointer after node failure, so a naive e‑commerce checkout runs like a transaction and the Timeline demo becomes practical.
AI-written summary · about
chapters (21)
- 0:00Distributed systems intro
- 0:23Why systems feel esoteric
- 3:11Centralized vs distributed
- 5:35Why systems are common
- 6:18Core challenges of state
- 9:28Naive updates fail
- 12:21The consensus problem
- 16:55CAP theorem limits
- 17:59Introducing Golem
- 22:40WebAssembly components
- 24:48Simplifying serverless
- 26:14Workers and reliable execution
- 29:25Golem’s superpowers
- 31:09Recovering from failures
- 35:23Does Golem solve state?
- 37:52Golem’s impact
- 40:00Building SQL with Golem
- 42:06Timeline project queries
- 43:32Becoming a wizard
- 43:53Systems eating the world
- 44:58Closing & resources
related talks
28mEnter a new world of web development where everything is serverless
50mWebAssembly: A Recovering Kubernetes Engineer's View of the Future
04mFeynman Liang Interview
29mOrchestrating Wasm: Reconciliation Loops Aren't Owned by Kubernetes
28mBOSH: The Simple Solution for Complex Systems
33mKubernetes on the edge: running low-latency workloads with WebAssembly
32mThe Go Circuit: Towards Elastic Computation with No Failures
13mRunning K8s in the Browser. Yes, Really. Well, Not Really. Kind of.