Adaptive Query Optimization and Workload-Aware Resource Allocation in CockroachDB Distributed Cloud Databases

Abstract

CockroachDB distributed cloud database solutions improve query processing and resource utilization by adaptive query optimization and workload-aware resource allocation. This technique uses workload detection, rule-based query plan selection, index evaluation, and threshold-based CPU, memory, and concurrency constraints. Contrary to traditional static configurations, this approach combines query optimization with resource management and relies on feedback during execution for adjustment of the following workload decisions regardless of artificial intelligence or machine learning techniques. The innovation consists in the integration of workload-aware query optimization with resource management in one distributed SQL optimization framework. The experimental validation is based on the Transaction Processing Performance Council Benchmark C (TPC-C) transactional workload that simulates wholesale transactions with nine relational tables, such as WAREHOUSE, CUSTOMER, ORDERS, and STOCK. Provided benchmark tables show illustrative results: total query latency varies from 32 to 90 ms across five workload classes, transaction throughput grows from 900 to 3,390 transactions per second. Meanwhile, CPU utilization goes from 22% to 60%, in contrast to 28% to 81% in default configuration. These preliminary numbers indicate potential performance gains, but controlled CockroachDB experiments should be run to prove the suggested optimization approach.