BigQuery, bytes counted before you run.
BigQuery REST v2 and GoogleSQL with dry-run cost inspection, per-query cost ceilings, typed result paging, schema tooling, IAM plan review, and encrypted logical archives.
GoogleSQL · service account / ADC · emulator
The dry run Mongrel shows before you spend.
Click any item below to see bytes processed, cost ceilings, and the metadata Mongrel surfaces. Same controls you get inside the workbench.
// Dry Run · bytes processed
Validates the GoogleSQL and reports total bytes processed without running the query. 412.6 GB against this profile's 100 GB maximum-bytes-billed ceiling - the job is rejected before a single byte is billed.
-- analytics · project acme-prod SELECT DATE(e.event_ts) AS day, u.country, COUNT(*) AS events FROM analytics.events e JOIN analytics.dim_users u USING (user_id) WHERE e.event_type = 'purchase' GROUP BY day, country ORDER BY day DESC; -- dry run: 412.6 GB processed · over the 100 GB bytes-billed ceiling -- add a partition filter and scan weeks, not years: WHERE _PARTITIONDATE >= '2026-07-01' AND e.event_type = 'purchase' -- dry run with the filter: 8.4 GB · 98% fewer bytes
What Mongrel ships for BigQuery.
GoogleSQL workbench
Run parameterized GoogleSQL with read/write classification, page-token results, and caller-owned job cancellation. Sessions give explicit BEGIN/COMMIT/ROLLBACK transactions.
Dry-run cost inspector
Run a dry run before any query. BigQuery reports the exact bytes processed up front, so runaway scans are caught before a single byte is billed.
Cost ceilings
Set a per-query maximum bytes billed ceiling - 100 GB by default. Every interactive query and dry run carries it, so runaway scans are stopped by the ceiling, not discovered on the invoice.
Flexible authentication
Service account JSON, OAuth access token, or Application Default Credentials. Point at the BigQuery emulator for local development.
Typed result paging
Strongly typed rows over BigQuery REST v2 with page-token paging. INT64, NUMERIC, GEOGRAPHY, and STRUCT arrive as real types.
Data movement
Export tables and query results to CSV, JSON, JSON Lines, or replayable INSERT statements, load local files into tables, and copy data between datasets and projects.
Schema tooling & drift
Browse datasets, tables, and views with partition and clustering metadata. Compare schemas across projects and detect drift.
IAM plan review
Inspect dataset access entries, then preview GRANT and REVOKE changes for the four supported dataset roles as reviewable plans before anything is applied to the project.
Encrypted logical archives
Snapshot tables, views, and routines into AES-256-encrypted logical archives. Restore into BigQuery with collision checks and every archived row preflighted.