cloud-cost-sql (MVP: AWS)

SQL-first cloud cost analytics you can run locally in minutes.

Why? Native tools are UI-heavy and hard to customize. This repo gives you drop-in SQL views over AWS Cost & Usage Report (CUR)-style data so engineers and analysts can query costs directly.

Quickstart

Requirements: Docker

git clone <your-fork-url>
cd cloud-cost-sql
docker-compose up -d
# wait ~10s for init, then connect to Postgres
psql -h localhost -U cloud -d cloudcost
# password: cloudpw

Now try some queries:

-- Daily total cost
SELECT * FROM aws.daily_trend;

-- Cost by service per day
SELECT * FROM aws.cost_by_service WHERE usage_day >= CURRENT_DATE - 30;

-- Top accounts by spend
SELECT * FROM aws.top_accounts;

-- Potential anomalies (spikes based on 7-day rolling z-score)
SELECT * FROM aws.anomalies ORDER BY usage_day DESC;

What’s Included

• Sample dataset (synthetic, small) that mimics AWS CUR columns
• Postgres schema (aws.cur) auto-loaded via Docker init
• Reusable views:
  • aws.cost_by_service
  • aws.daily_trend
  • aws.anomalies (rolling 7-day z-score)
  • aws.top_accounts

Roadmap

• GCP (BigQuery export) + Azure (Cost Management exports)
• dbt packages for warehouse-native installs
• Prebuilt Power BI / Superset / Metabase dashboards
• FinOps-friendly dimensions (tags, savings plans, RI coverage)

Troubleshooting

I don’t see the views (aws.daily_trend, etc.)
Postgres runs files in docker-entrypoint-initdb.d only on first init of the data volume. If your container initialized before the views file existed, do one of the following:

• Apply views without data loss:

  ./scripts/apply_views.sh
  # or: make apply-views
  
  

Contributing

Issues and PRs welcome! Please open an issue to discuss changes before large PRs.

License

MIT