Connection requirements
- A GCP service account JSON key with BigQuery Data Editor on the target dataset
- An existing (or NoSQLSync-created) dataset in the project
Streaming insert, then MERGE
Loading into an empty table uses BigQuery's streaming insert API (table.insert) — no DML overhead. Once the table has rows, both the bulk load and CDC use a MERGE statement so re-runs and live updates stay idempotent.
Setup
1
Add the connection
Provide the BigQuery connection details under Connections → Add. NoSQLSync verifies it can connect before saving.
2
Pick a table name
Choose an existing table or let NoSQLSync create one from the inferred schema when the migration starts.
3
Start the migration
Bulk-load the historical data, then optionally enable CDC to keep the table current.
FAQ
Does NoSQLSync create the BigQuery table for me?
Yes. If the target table doesn't exist, it's created from the inferred schema before the first batch loads.
What happens if I re-run a migration that already has data in the target?
The bulk load switches to an idempotent upsert path instead of the empty-table fast path, so re-running is safe.