Skip to main content

Architecture

ReifyDB is built with a modern, modular architecture designed for performance and scalability.

System Overview​

ReifyDB consists of several key components:

  • Query Engine: Parses and optimizes SQL queries
  • Storage Engine: Manages data persistence with MVCC
  • Transaction Manager: Handles ACID transactions
  • Network Layer: Provides gRPC and WebSocket interfaces
  • Flow Engine: Manages real-time data flows and CDC

Storage Architecture​

Multi-Version Concurrency Control (MVCC)​

ReifyDB uses MVCC to provide high concurrency without locking:

  • Each transaction sees a consistent snapshot of the database
  • Writers don't block readers
  • Readers don't block writers

Storage Backends​

ReifyDB supports multiple storage backends:

  • Memory: In-memory storage for development
  • LMDB: Lightning Memory-Mapped Database for production
  • SQLite: Embedded SQL database for single-node deployments

Query Processing​

  1. Parse: SQL is parsed into an Abstract Syntax Tree (AST)
  2. Analyze: Type checking and validation
  3. Optimize: Query optimization and plan generation
  4. Execute: Physical plan execution

Transaction Model​

ReifyDB supports:

  • ACID Guarantees: Full ACID compliance
  • Isolation Levels: Read Committed, Repeatable Read, Serializable
  • Optimistic Concurrency: Conflict detection at commit time

Network Protocols​

  • gRPC: High-performance RPC for applications
  • WebSocket: Real-time bidirectional communication
  • HTTP/REST: RESTful API for web applications