Energy-efficient workflow scheduling using dynamic task clustering for sustainable cloud computing

In the dynamic landscape of cloud computing, efficient scientific workflow scheduling remains a significant challenge. This research introduces a pioneering approach that combines dynamic task clustering and sustainability considerations. The core objective is to optimize task execution while achieving resource consolidation. Leveraging Directed Acyclic Graphs (DAGs) as workflow representations, the approach incorporates the Dynamic Cloud Task Clustering with Optimization and Scheduling (DCTCOS) algorithm. By efficiently identifying and grouping related tasks into clusters, scheduling overhead for fine-grained scientific tasks is reduced, fostering improved parallel processing in dynamic cloud environments. Notably, the proposed DCTCOS algorithm demonstrates performance enhancement ranging from 2 to 19 percentage points compared to existing benchmark algorithms, depending on the workflow and baseline for comparison. Simultaneously, the study addresses the sustainability challenges cloud data centers face due to escalating energy consumption. The Green-Aware Workload Shifting Algorithm strategically reallocates virtual machines (VMs) and integrates solar energy sources, optimizing resource utilization and mitigating environmental impact. Rigorous experiments consistently validate the effectiveness of this approach, promising prospects for more efficient and resource-conscious scientific computing practices.