SGA: Intelligent Task Scheduling for Enhanced Throughput and Fault Tolerance in Cloud Environments

Task scheduling in cloud computing plays a crucial role in achieving high reliability and performance under dynamic and heterogeneous workload conditions. However, existing scheduling approaches often suffer from high Virtual Machine (VM) failure rates and limited throughput due to inefficient load distribution and poor adaptability. To address these issues, this study proposes a bio-inspired Shrimp and Goby Association Search (SGA) algorithm for reliable and performance-aware task-VM mapping in cloud environments. The proposed scheduler is implemented in the CloudSim simulation framework and evaluated using realistic Google Cloud workload traces, with comparative analysis against Ant Colony Optimization (ACO), Particle Swarm Optimization (PSO), and Whale Optimization Algorithm (WOA). Experimental results clearly demonstrate the effectiveness of SGA, achieving up to 19.6% reduction in VM failure rate, 14.8% improvement in task success rate, and 17.3% increase in system throughput compared to the benchmark algorithms. These improvements are achieved through the cooperative exploration-exploitation mechanism of SGA, which ensures balanced load distribution and adaptive scheduling under varying workloads. The results confirm that the proposed SGA-based scheduler provides a robust, scalable, and reliable solution for large-scale cloud infrastructures, significantly enhancing quality of service and overall resource utilization.