KaiwuDB Docs

Download PDFSubmit Feedback

Product Architecture

KWDB takes a different approach from traditional databases by offering multi-model data management, allowing you to store various types of data in a single, unified system. This flexibility enables businesses to seamlessly integrate and manage data from different departments and applications, making it possible to support a wide range of services under one roof.

The architecture of KWDB is illustrated below:

KWDB Architecture

Interface Layer: Supports multiple standardized interface protocols, including JDBC, ODBC, and RESTful API, ensuring seamless compatibility with various applications and development tools.

Computing Layer:

  • SQL Execution Engine: Delivers a complete SQL processing workflow, including protocol parsing, multi-model SQL parser, multi-model SQL optimizer, and multi-model SQL executor. Enables unified querying and processing across time-series and relational data.

  • Monitoring and Management Module: Provides comprehensive system monitoring and management capabilities, including session management, memory management, thread management, message management, heartbeat management, performance monitoring, and log tracing.

  • Core Service Components: Includes essential database services such as metadata management, transaction management, job management, and security management.

  • Value-Added Services: Offers advanced features including real-time compression, lifecycle management, specialized built-in functions, system tasks, and Write-Ahead Logging (WAL).

  • Multi-Engine Integration: Integrates multiple data computing engines that automatically select the optimal storage and computation mode based on data characteristics while providing a unified interface. This delivers diverse analytical capabilities and enhanced query efficiency.

    • Adaptive Time-Series Engine: Specifically optimized for time-series data, supporting complex time-series queries and multi-dimensional aggregation. Capable of inserting millions of records in real-time within seconds and delivering query responses in seconds for hundreds of millions of records. Provides 5-30x real-time compression, with compressed data directly usable without decompression.
    • Transaction Processing Engine: Supports distributed transactions and Multi-Version Concurrency Control (MVCC), with complete relational database capabilities including comments, views, constraints, indexes, sequences, stored procedures, and triggers.

Storage Layer:

  • Time-Series Storage Engine: Optimized for time-series data, enabling efficient data storage and retrieval.
  • Relational Storage Engine: Traditional relational database engine that provides full ACID transaction guarantees.