ServicesAboutNotesContact Get in touch →
EN FR
Note

BigQuery Slots and Reservations

A reading guide to BigQuery's compute model -- slots, reservations, editions, autoscaling, fair scheduling, and slot management for dbt workflows.

Planted
bigquerydbtcost optimization

This hub covers BigQuery’s compute model: slots, reservations, editions, autoscaling, fair scheduling, and slot management for dbt workflows. Notes build on each other in sequence — start with slot fundamentals, then capacity organization, then monitoring and dbt-specific guidance.

Reading Order

BigQuery Slots — the fundamental compute unit; everything else builds on this concept.

BigQuery Reservation Hierarchy — how capacity is organized: commitments buy slots, reservations create isolated pools, assignments route projects to those pools.

BigQuery Editions — the three pricing tiers (Standard, Enterprise, Enterprise Plus) and their feature differences relative to on-demand.

Baseline vs. Autoscaling Slots in BigQuery — guaranteed baseline vs. elastic autoscaling within a reservation, including the 60-second autoscale billing window.

BigQuery Fair Scheduling — how slots are distributed when multiple queries compete, and why project architecture affects query performance.

BigQuery Idle Slot Sharing — Enterprise’s mechanism for borrowing unused slots across reservations, with preemption trade-offs.

dbt Slot Management on BigQuery — applying slots and reservations to dbt workflows: compute characteristics, multi-project reservation patterns, and sizing guidance.

BigQuery Slot Usage Monitoring — queries and tools for understanding actual slot consumption before making capacity decisions.

BigQuery Autoscaling Cost Overhead — why real-world Editions costs exceed theoretical estimates: the 1.5x multiplier, the 60-second billing window, and workload shape effects.

BigQuery Editions Testing Without Commitment — creating a test reservation, using the none reservation to opt projects out of org-level configurations, and the rollback procedure.

BigQuery Editions Migration Anti-Patterns — five consistent mistakes when migrating from on-demand: over-provisioning max slots, ignoring the billing window, skipping query optimization, purchasing commitments before testing, and consolidating all workloads into one reservation.

Prerequisites

Familiarity with how BigQuery works under the hood — storage/compute separation, columnar storage, the Dremel execution engine — and the BigQuery Cost Model pricing math.

BigQuery SQL Patterns for Analytics Engineers — query-level optimization. Slots determine available compute; SQL patterns determine how efficiently it is used.