To SQL or NoSQL? Firebase is the Answer
summary
Bunny Search uses Firebase DataConnect (a SQL layer introduced 2024) to run a Postgres instance on Cloud SQL with a many‑to‑many brand↔collection schema and a compound primary key; GraphQL schemas generate type‑safe SDKs but require writing raw SQL for a Postgres trigram extension fuzzy search and JSON blob views. DataConnect is paid (Blaze, ~\$9+ mo, three free months), reached GA under six months, drops indices on some compatible deploys, has no real‑time updates, and once needed a redeploy to restore production.
AI-written summary · about
chapters (17)
- 0:00Firebase Data Connect intro
- 1:23Bunny Search data model
- 3:52Relational database choice
- 6:20Data Connect setup
- 9:33GraphQL with Data Connect
- 11:39GraphQL schema setup
- 14:24CRUD in GraphQL
- 17:59Auth and security
- 20:31Client app connection
- 22:39Testing and deployment
- 24:03Schema migration
- 26:52Advanced Data Connect
- 27:33Fuzzy search in PostgreSQL
- 29:00JSON fields handling
- 30:45AI features in Data Connect
- 31:49Should you use Data Connect?
- 33:15Data Connect challenges
related talks
23mFANTASTIC DATABASES AND WHERE TO FIND THEM
26mHow to select a future-proof database
47mFrom SQL to NoSQL
46mPower the next generation of AI apps with databases at scale anywhere
21mUse cases that drive the need for NoSQL databases
01h 02mDiamond Sponsor Hall
37mBuilding Scalable Apps with Cloud Firestore
53mThe NoSQL Store everyone ignores: PostgreSQL