Concepts

This section explains how VaireDB works — what each node does, how data is distributed and replicated, how queries run, and what guarantees you can rely on.

• Design Goals

  What VaireDB optimizes for, and what is explicitly out of scope for v0.1.

• System Architecture

  The coordinator/core topology and node responsibilities.

• Coordinator Node

  Query routing, distributed planning, and the metadata catalog.

• Core Node

  The embedded DuckDB engine, storage, and query execution.

• Data Distribution

  Hash sharding and quorum-based replication.

• Query Processing

  The read, write, and DDL paths, end to end.

• Cluster Coordination

  Node discovery, primary assignment, and failure detection.

• Communication Layer

  Wire protocols between clients, the coordinator, and core nodes.

• Transactions & Consistency

  The per-shard consistency model and its limitations.

• Fault Tolerance

  WAL, quorum, recovery, and planned snapshotting.