How to Migrate from Firebase to PandaStack

Firebase is a fantastic place to start: Hosting, Cloud Functions, Firestore, and Auth in one console. The friction shows up later — when you want a relational database, predictable pricing on a real backend, or a container you fully control. This guide walks through migrating a typical Firebase app to PandaStack, component by component, and is honest about the parts that require genuine re-architecture.

What maps to what

The two genuinely hard parts are Firestore → relational and Firebase Auth. Everything else is mostly mechanical. Be honest with yourself about scope before you start.

Step 1: Migrate Hosting (the easy win)

If your Firebase Hosting site is a static SPA, this is the simplest piece. PandaStack auto-detects frameworks (React/Vite, Next export, Astro, Gatsby, Eleventy, VitePoint, Hugo, plain HTML) and runs the build for you.

1. 1Connect your Git repo in the PandaStack dashboard.
2. 2Confirm the detected build command and output directory (override if needed — install command supports npm/yarn/pnpm/bun).
3. 3Deploy. Static builds run in pandastack.ai microVMs.
4. 4Point your custom domain; SSL is issued automatically.

If your firebase.json had rewrites for an SPA fallback, replicate that with a single-page-app fallback to index.html.

Step 2: Decide on your database model

This is the decision that shapes the whole migration. Firestore is a document store; PandaStack offers managed MongoDB (closest to Firestore's document model) and managed PostgreSQL/MySQL (relational).

• Lift-and-shift: choose MongoDB. Your document shapes transfer with minimal change. Fastest path.
• Re-architect to relational: choose PostgreSQL. More upfront work modeling tables and relationships, but you gain joins, transactions, constraints, and SQL — often the reason people leave Firestore in the first place.

A pragmatic middle path: lift to MongoDB now to unblock the migration, then move hot collections to PostgreSQL later, one at a time.

Step 3: Export Firestore data

Firestore exports to Cloud Storage in its own format, which isn't directly importable elsewhere. The reliable approach is a script using the Admin SDK to read collections and write portable JSON:

// export-firestore.js
import admin from 'firebase-admin';
import { writeFileSync } from 'fs';

admin.initializeApp({ credential: admin.credential.applicationDefault() });
const db = admin.firestore();

const snap = await db.collection('users').get();
const docs = snap.docs.map(d => ({ id: d.id, ...d.data() }));
writeFileSync('users.json', JSON.stringify(docs, null, 2));
console.log(`Exported ${docs.length} docs`);

Watch out for Firestore-specific types: Timestamp, GeoPoint, DocumentReference. Convert timestamps to ISO 8601 strings and references to plain IDs during export.

Step 4: Import into your managed database

Provision the database in PandaStack. When you attach it to your app, the connection string is injected automatically as DATABASE_URL — no copying secrets between dashboards.

For MongoDB:

mongoimport --uri "$DATABASE_URL" --collection users --file users.json --jsonArray

For PostgreSQL, write a small loader that maps each JSON document to a row, flattening nested fields into columns or jsonb as appropriate:

CREATE TABLE users (
  id        TEXT PRIMARY KEY,
  email     TEXT,
  created_at TIMESTAMPTZ,
  profile   JSONB
);

jsonb is a useful escape hatch for irregular nested data while you incrementally normalize.

Step 5: Migrate Cloud Functions

Triage your functions into two buckets:

• Short, stateless, event-ish (webhook handlers, lightweight transforms): port to edge functions (Node.js, Python, Go runtimes).
• Long-running, stateful, or framework-heavy (an Express API, background processing): fold into a container app. Often it's cleaner to consolidate a dozen HTTP-triggered functions into one container running your existing framework.

The big behavioral change: Firebase Functions auto-injected Firebase Admin and Firestore access. After migration you connect to your database via DATABASE_URL like any normal app. That's more explicit — and more portable.

Step 6: Replace Firebase Auth (the honest part)

There is no drop-in replacement for Firebase Auth, and pretending otherwise wastes your time. Options:

• Run an open-source auth server as a container app.
• Use a third-party auth provider via integration.
• Build minimal JWT auth if your needs are simple.

If you have existing users, plan a password-reset-on-first-login flow — you generally cannot export Firebase password hashes in a directly reusable form for arbitrary systems. Migrate user records (email, profile, roles); force a credential reset for the secret material.

Step 7: Cutover

1. 1Deploy the full stack (static + container/functions + DB) on PandaStack and test against a staging domain.
2. 2Do a final delta export/import for data written since your first import (or run dual-write briefly).
3. 3Switch DNS to PandaStack. SSL auto-provisions.
4. 4Keep Firebase running read-only for a rollback window before tearing it down.

Migration checklist

• [ ] Hosting → static site deployed and domain tested
• [ ] Database model chosen (MongoDB vs PostgreSQL)
• [ ] Data exported with type conversions handled
• [ ] Data imported; row/doc counts verified
• [ ] Functions triaged and ported
• [ ] Auth replacement chosen and tested
• [ ] Delta sync + DNS cutover plan written
• [ ] Firebase kept as rollback for a window

References

• [Firebase Admin SDK](https://firebase.google.com/docs/admin/setup)
• [Firestore data export/import](https://firebase.google.com/docs/firestore/manage-data/export-import)
• [MongoDB mongoimport](https://www.mongodb.com/docs/database-tools/mongoimport/)
• [PostgreSQL jsonb type](https://www.postgresql.org/docs/current/datatype-json.html)
• [Twelve-Factor App: backing services](https://12factor.net/backing-services)

---

The nicest part of landing on PandaStack: connect your repo, and your managed database is auto-wired with DATABASE_URL injected — no glue code. The free tier (1 database, 5 web services, edge functions included) is enough to run the whole migration as a dry run. Start at [dashboard.pandastack.io](https://dashboard.pandastack.io).