Autonomic Lock Escalation in SQL Database Queries

The patent implements dynamic lock escalation by monitoring the number of locks acquired during query execution and automatically transitioning from row-level locking to page-level locking when a threshold is reached. This dynamic adjustment allows the system to adapt to changing concurrency conditions, maintaining high concurrency when needed while switching to more efficient page-level locking when the number of locked rows exceeds a predefined threshold, thus resolving the contradiction between concurrency and execution efficiency.

The patent applies preliminary action by pre-configuring lock escalation thresholds and policies before query execution begins. The system monitors lock acquisition in real-time and triggers escalation proactively when predetermined conditions are met, rather than waiting for performance degradation to occur. This advance planning and monitoring mechanism ensures that the transition to page-level locking happens at the optimal moment, reducing system overhead before performance issues manifest.

The patent implements segmentation by dividing the locking strategy into two distinct levels: row-level locking for initial query execution to maintain low contention, and page-level locking for bulk operations where efficiency is paramount. The system segments the decision-making process by evaluating query characteristics and lock count thresholds, applying the appropriate locking granularity for each scenario. This segmented approach allows the system to leverage the advantages of both locking strategies while minimizing their respective disadvantages.