"How We Built a Rust-y Real-Time Public Transport Map"
summary
A Rust real-time map that pulls GTFS-RT from 800 agencies at one-second cadence, uses etcd+Raft to assign agency shards to workers, and serves tile and vehicle queries over Actix Web. Global polling would be about 10 TB/day, so precompute transfer patterns (Hannah Bass/BAS-style), compress schedules from 600GB→100GB, and partition OSM with K-means. Uses SQLx/Diesel, Tokio and Rayon for concurrency, OSM PBF for geometry, and runs acceptably even on older Intel Nooks; demo works live.
AI-written summary · about
chapters (12)
- 0:00Project motivations
- 1:17Demo
- 2:26Agency URLs
- 3:41Distributed Systems
- 5:04GTFS
- 6:10Postgress
- 7:40RealTime Database
- 9:48Algorithm
- 12:38Compression
- 14:10Preprocessing
- 16:20Query Time
- 16:59Excerpt
related talks
28mPowering Crowd-sourcing of Public Transit Delays with Rust
55mScaling out operational research with Rust
46mAdopting Rust by gracefully oxidising web applications
35mAll aboard the Rust (electric freight) train!
47mRust At The Edge
38mTokio, Rayon, and the GPU: Profiling a Radiology AI Pipeline in Rust
23mRoverMap: Rendering a Vector Map With Rust and Vulkan
42mBuilding a Web Application with Rust