Design a planet-scale Quota Management System that enforces strongly-consistent resource limits for every customer of Snowflake’s cloud data platform. The service must track and police multiple quota dimensions per organization: warehouse compute credits per hour/day/month, storage bytes, concurrent queries, API rate-limits, and data-transfer egress. Quota rules are hierarchical: organization → account → warehouse → user, with inheritance and optional burst allowances. All backend services (query optimizers, loaders, serverless tasks) must atomically reserve quota before starting work; if none is available the request must be rejected within 10 ms at p99. The system should sustain 10 M reservations/s globally, tolerate region failure without over-allocation, and provide millisecond-accurate usage dashboards to end-users. Outline the APIs, data model, reservation protocol, caching strategy, failure handling, and rollout plan.