Architecture

High-Level Flow

1. Client connects to gateway on MongoDB wire protocol.
2. Gateway authenticates and dispatches commands.
3. Coordinator decides hot, cold, or federated execution.
4. Storage Engine and/or Query Engine execute.
5. Coordinator merges and returns a MongoDB-compatible response.

Request Ownership

Request Type	Primary Owner
Handshake, auth, command parsing	Gateway
Query routing/merge/cursor orchestration	Coordinator
Hot writes and canonical query/update semantics	Storage Engine
Cold object-store scans and pushdown	Query Engine
Lifecycle archival + backup/restore orchestration	Controller

Primary Code Anchors

• Gateway dispatch: services/gateway/src/proxy/handler/mod.rs
• Gateway command catalog: services/gateway/src/proxy/handler/handshake.rs
• Coordinator orchestration: services/coordinator/internal/server/server.go
• Storage aggregation/filter/update execution: services/storage-engine/src/aggregation/pipeline.rs, services/storage-engine/src/filter/evaluator.rs
• Query-engine cold execution: services/query-engine/src/grpc/service.rs

Operational Topologies

• Local single-node: docker-compose.yml
• Local cluster E2E: docker-compose.yml + docker-compose.e2e.yml
• Kubernetes/Cloud: deploy/helm, deploy/kustomize, deploy/terraform

Verification Gates

• Unit/integration/compatibility: make test, make test-integration, make test-compatibility
• E2E topologies: make e2e-single, make e2e-local-cluster
• MQL matrix: make mql-exhaustive, make mql-coverage-audit