Open Source Routing Machine

Open Source Routing Machine (OSRM) is a high‑performance routing engine written in modern C++ that ingests OpenStreetMap (OSM) data and exposes a lightweight HTTP API, a C++ library interface, and a Node.js wrapper. At its core, OSRM implements state‑of‑the‑art shortest‑path algorithms—Contraction Hierarchies (CH) and Multi‑Level Dijkstra (MLD)—to compute routes, distance matrices, and map‑matching in real time. The engine is designed for self‑hosting; Docker images are provided for rapid deployment, but the binaries can also be compiled from source and run on any Debian‑based host.

Key Features

• Routing Services: route, nearest, table, match, trip, and tile endpoints allow developers to retrieve fastest routes, snap coordinates to the road network, generate distance/duration matrices, perform map‑matching of noisy GPS traces, solve TSP heuristics, and produce Mapbox Vector Tiles enriched with routing metadata.
• Customizable Profiles: A user‑defined routing profile (written in Lua) specifies vehicle types, speed limits, turn restrictions, and penalties, enabling fine‑grained control over travel behavior for cars, bicycles, pedestrians, or any custom mode.
• Extensible API: The C++ libosrm library can be embedded into other applications, while the Node.js wrapper exposes the same functionality to JavaScript developers. The HTTP API follows a REST‑like contract documented in docs/http.md, making it trivial to integrate with existing web stacks.

Technical Stack & Architecture

• Language: C++17/20 for core engine; Lua for routing profiles.
• Data Structures: Binary routing graphs serialized to disk using custom formats (e.g., osrm-*.osrm). Pre‑processing steps (osrm-partition, osrm-customize, osrm-contract) generate auxiliary data (hierarchy levels, contraction hierarchies) that accelerate query time.
• Algorithms:
  • Contraction Hierarchies (CH) – fast for large distance matrices, single‑source shortest paths.
  • Multi‑Level Dijkstra (MLD) – efficient for general routing queries, scalable to continental‑sized networks.
• No External DB: All data is loaded from pre‑processed files into memory; no external database required, simplifying deployment.

Deployment & Infrastructure

OSRM is stateless once the routing data is loaded; it can be horizontally scaled by running multiple instances behind a load balancer. Docker images are lightweight (Debian base, ~200 MB) and support docker‑compose setups for the backend plus optional frontend (osrm-frontend). For high availability, developers can mount the .osrm files on a shared NFS or use container orchestration (Kubernetes) with persistent volumes. Memory consumption scales linearly with network size—typical continental datasets fit comfortably in 32–64 GB of RAM.

Integration & Extensibility

• Plugins: While OSRM itself has a limited plugin system, custom routing profiles and Lua scripts act as plug‑ins to alter behavior without modifying the core code.
• Webhooks & Callbacks: The HTTP API supports JSON payloads; developers can hook into route calculation events via middleware in their own services.
• Frontends: The open‑source osrm-frontend provides a Mapbox GL JS interface, but any JavaScript mapping library can consume the API. Text instructions are generated by osrm-text-instructions.

Developer Experience

• Documentation: Comprehensive docs are hosted at project-osrm.org and include API reference, profile authoring guides, and pre‑processing tutorials. The source repo contains a detailed docs/ directory.
• Community: Active Discord, IRC, and mailing list channels provide quick support. The project’s CI pipeline ensures that new commits are rigorously tested across multiple OSes.
• Licensing: BSD‑3 license grants full freedom to modify and redistribute, making OSRM attractive for commercial deployments.

Use Cases

• Fleet Management: Real‑time routing and distance matrices for delivery fleets, with custom vehicle profiles.
• Navigation Apps: Embedding libosrm in mobile or embedded devices for offline navigation.
• Geospatial Analytics: Generating origin‑destination matrices at scale for urban planning or traffic simulation.
• Custom Transport Modes: Defining unique routing logic (e.g., electric vehicle range constraints) via Lua profiles.

Advantages Over Alternatives

• Performance: CH and MLD deliver sub‑millisecond route queries on continental datasets, outperforming many commercial APIs.
• Flexibility: Full control over data ingestion and profile logic; no vendor lock‑in.
• Scalability: Stateless architecture allows easy horizontal scaling and integration with cloud-native stacks.
• Open Source & Free: BSD license removes cost barriers, while community support ensures rapid bug fixes and feature additions.

OSRM thus offers developers a robust, high‑throughput routing engine that can be tailored to any transportation scenario and deployed entirely on their own infrastructure.