Adventures in FHIR Resource Syncing and Caching
summary
Wellsheet moves FHIR fetch logic into a relational SQL cache and ingests resource fields rather than storing raw JSON, then builds a dependency graph so fetches run by level not waterfall, cutting a seven‑step sync to three parallelized tree layers and nearly halving time to first view. Cache state is tracked per resource with synced_from / synced_until timelines to support both older and newer queries. Explored optimizations include _include / _revinclude and FHIR GraphQL; multi‑EHR caches remain segregated.
AI-written summary · about
chapters (7)
- 0:00Introduction
- 1:03Direct FHIR API fetching
- 3:33Caching in SQL database
- 7:23Cache invalidation challenges
- 10:12View-specific syncing strategies
- 13:23Summary and key takeaways
- 14:25Q&A session
