dbt model SLA breach: model finishes 4 hours past SLA

Ситуация: Critical dbt model "user_metrics_daily" supposed to finish by 6 AM. Last week — 10 AM. Dashboards lag. Why?

dbt scheduled 3 AM. Reads from raw → transforms → writes user_metrics_daily. Used by 20 dashboards.

Доступные данные

• dbt_runs: run_id, ts, model, duration, status
• dbt_dag: model, upstream_dependencies
• data_volume: table, daily_rows, growth_rate

Задачи

• Run timing decomposition.
• Bottleneck: which model slowest.
• Volume growth.
• Solutions: optimize, materialize, partition.

Все кейсы для подготовки → ← Все кейсы