Waterpark: Transforming Healthcare with Distributed Actors
summary
Waterpark is a BEAM/Elixir platform at HCA Healthcare that runs as a RAM-based database with zero downtime in production since 2020, modeling millions of patient actors as long‑lived digital twins. It uses rendezvous hashing, process pairs with read replicas in multiple data centers, and quorum writes (ack from two of four readers) so messages never lose. A runtime plugin system pushed a COVID nursing‑home alert in weeks and demonstrably reduced outbreaks.
AI-written summary · about
chapters (15)
- 0:00Introduction to Waterpark
- 3:07Goals and architectural vision
- 7:02Platform implementation
- 9:34Hardware and system choices
- 12:22The actor model in detail
- 17:05Patient-centric data modelling
- 20:42HL7 data and bloom filters
- 24:54Continuous availability
- 27:57Fault tolerance and resilience
- 30:21Routing and topology management
- 34:04Consensus and leader election
- 37:46Location transparency and mailrooms
- 39:29Invariants and deployment strategies
- 42:32Universal servers and plugins
- 45:48System rewrite and conclusion
related talks
41mAdopting Microservices in Healthcare, Breaking Apart the Monolith
13mBuilding the next-gen healthcare platform using cloud native bingo
38mSystems playing nice: Building resilient integrations in an unpredictable world
27m"Write Once Stream Everywhere: Transforming Your Data into Events"
39mRethinking RDBMS Data Migration
27mScalable distributed systems from grounds up in Elixir
34mSWIMming in the Microservices Ocean
28mSWIMming in the microservices ocean