Scaling beyond Switzerland: the Alps data pipeline

TrailSplits started Switzerland-first for speed and iteration. The next step is expanding to the Alps while keeping the same operational discipline: repeatable builds, sanity checks, and safe activation.

The contract: releases + activation

Instead of overwriting production assets in-place, TrailSplits treats data outputs as immutable releases and then activates them safely.

At a high level:

  1. Build artifacts on a compute box (Server 2)
  2. Ship artifacts to the serving box (Server 1)
  3. Validate endpoints and metadata
  4. Attach datasets into the active release (without breaking the live map)

This release discipline makes it possible to scale regions without “hope-based deploys.”

What gets built

For the Alps expansion, the pipeline produces:

  • TerrainRGB DEM tiles (for elevation + terrain)
  • Terrain-intel overlays (slope/aspect/hillshade/ruggedness)
  • Contours vector tiles
  • Search index
  • Routing tiles

Each artifact is publishable independently, but the goal is a one-command refresh.

Why this matters

  • Credibility: data is reproducible and traceable to a release
  • Reliability: production is updated only after validation passes
  • Scalability: new regions are “just another release”

If you’re curious about the public-facing endpoints these datasets power, see /api.

Open Planner More articles