Intelligent load balancing and concurrency control in cloud-based distributed databases: A machine learning approach

Cloud-based distributed databases are critical for scalable modern applications, yet they struggle with uneven resource utilization and transaction conflicts. This paper introduces a machine learning (ML)driven framework combining reinforcement learning (RL) for dynamic load balancing and a hybrid concurrency control protocol. The RL agent optimizes query distribution by analyzing real-time node metrics, while the concurrency controller adaptively switches between optimistic and pessimistic strategies based on conflict predictions. Evaluations on AWS EC2 using the YCSB benchmark demonstrate a 30% improvement in throughput, 25% reduction in latency, and a 47% decrease in abort rates compared to traditional methods. The results validate the efficacy of AI-driven solutions in enhancing cloud database performance and scalability.