Abstract:
The scalability of cloud services comes with performance limitations due to resource contention and inefficient task scheduling. This leading to increased latency, reduced throughput, poor resource utilization, higher operational costs, and degraded Quality of Service. This study presents an innovative approach by integrating dynamic task scheduling with Weighted Least Recently Used (WLRU) caching to address these issues. WLRU caching improves cache efficiency by prioritizing critical data, such as task execution details, virtual machine (VM) metrics, and VM failure rates at the data center level, reducing latency and enhancing system performance. The dynamic scheduling algorithm adapts to workload variability, optimizing resource utilization and task execution. The methodology integrates WLRU caching with the Heterogeneous Earliest Finish Time (HEFT) scheduling algorithm, which has not been considered in prior studies. Simulations were conducted using CloudSim, focusing on task completion time as the key performance metric across two test scenarios. The integrated approach outperforms traditional HEFT without caching, improving task completion time by 17.78% for small workloads (3 VMs with 1–5 GB RAM, 10 cloudlets with 4,000–20,000 million instructions) and 15.23% for large workloads (10 VMs with 1–5 GB RAM, 100 cloudlets with 4,000–20,000 million instructions). By storing detailed task execution data and VM metrics, this approach significantly decreases latency and improves resource usage. The study’s findings underscore the potential of combining caching and scheduling techniques to optimize cloud performance, offering a unique solution for addressing the limitations of current models and contributing to more efficient, scalable, and sustainable cloud systems. Future studies can incorporate multiple performance metrics, such as cache hit and miss ratio and power usage.
