Query Execution Time Variability in Oracle Autonomous Database Workloads

Abstract

Query execution time in Oracle Autonomous Database is influenced not only by SQL structure and data
characteristics but also by the platform’s elastic resource allocation mechanisms. This study examines
how execution time varies across different system load states, focusing on the transitional phases where
autoscaling increases or decreases compute capacity in response to concurrent workload demand.
Results show that execution times remain stable during low-load and fully scaled steady states, but
temporary latency increases occur during ramp-up and scale-down intervals. The variability is therefore
systematic and cyclical rather than random, reflecting the timing of autonomous resource adjustments.
Understanding these temporal elasticity patterns is essential for accurate performance evaluation,
capacity planning, and workload scheduling in real-world Autonomous Database environments.