Updated on 2025-09-04 GMT+08:00

Architecture Principles

Core Principles

  • Data type optimization: Integer > Floating-point number > Numeric (in order of priority).
  • Index balancing mechanism: trade-off between query acceleration and update costs.
  • Partition storage policy: logically unified + physically dispersed.
  • Storage engine features: Ustore supports in-place update, and Astore supports append-only.

Solution Advantages

  • Improved query performance: Index design and partitioning policies help reduce the scanned data size.
  • Reduced storage costs: Selecting appropriate data types can save more than 30% of space.
  • Enhanced maintenance efficiency: Independent partition maintenance minimizes impact on services.
  • Enhanced concurrent processing: Concurrent access to multiple partitions improves the service throughput.