Distributed Database Data Loading with Dual-Key Sorting

The patent applies preliminary action by sorting data records by key values before loading them into the distributed database. The system generates candidate files with pre-sorted data and pre-calculated group counts, selecting the optimal candidate file before the actual loading process. This pre-processing ensures even data distribution across storage resources without adding complexity during the loading phase itself.

The system applies self-service by automatically evaluating multiple candidate files and selecting the optimal one based on group count metrics. The database management system autonomously determines which candidate file will provide the most even data distribution, eliminating the need for manual configuration or external intervention. This self-optimizing approach ensures high searching efficiency without requiring additional time for manual data distribution tuning.

The patent applies partial action by generating multiple candidate files with different sorting configurations, but only evaluating them based on the critical metric of group count. The system generates sufficient candidate files to ensure optimal data distribution, then selectively evaluates them using a streamlined process that counts groups without performing exhaustive analysis. This approach achieves high data distribution quality while limiting processing time through focused evaluation criteria.