CONNECTOR · BIGQUERY
BigQuery — GCP-native warehouse. Pay-per-byte makes partition discipline non-negotiable.
CONNECTOR · BIGQUERY
BigQuery
GCP-native warehouse. Pay-per-byte makes partition discipline non-negotiable.
- categoryDestination · Warehouse
- freshness targetStreaming ingest available; we usually run minute-grain.
- auth modelService account via Workload Identity. Per-dataset IAM.
- sync modedbt incremental on partitioned tables. Streaming inserts where freshness demands.
TABLES SYNCED
Partitioned by ingestion or event date; clustered on customer / product keys.
SCHEMA · SAMPLE
schema.sqlSQL
CREATE TABLE marts.fct_events (
event_ts TIMESTAMP,
customer_id STRING,
event STRING,
amount NUMERIC
)
PARTITION BY DATE(event_ts)
CLUSTER BY customer_id, event;QUERY · SAMPLE
query.sqlSQL
SELECT event, COUNT(*) AS n
FROM marts.fct_events
WHERE DATE(event_ts) >= DATE_SUB(CURRENT_DATE, INTERVAL 30 DAY)
GROUP BY event
ORDER BY n DESC;OPS NOTES
Partition pruning is the difference between a $0.10 query and a $40 query. Every PR enforces a partition predicate on fact tables.
Wire a real pipeline to your warehouse
30-min stack review. We come back with a scoped plan, an owner, and a freshness contract. No SDR, no slide deck.