WORKLOADAWARE: A Resource Allocation and Consolidation Technique for Heterogeneous Clouds

Abstract

—Datacenters are the core of cloud computing services that consume a significant amount of energy and negatively affect our environment. These problems can be solved using efficient resource management techniques, such as resource allocation and consolidation methods. This paper models the resource allocation problem as a bin-packing NP-hard problem and proposes a workload-aware resource allocation and consolidation method that accounts for the heterogeneity of resources and applications. Using plausible assumptions and real datasets from Google cluster traces, we investigated the impact of workload-aware scheduling and migration methods on infrastructure energy efficiency and workload performance. We found that our suggested method can provide around 5.25% energy savings and 8.17% workload performance gains on 12,583 heterogeneous servers and over three million tasks across three distinct applications. Moreover, a workload-aware migration technique can lower energy consumption (6.42% – 18.3%) and increase performance (7.8% – 25.03%) for specific applications, despite the fact that we found an existing trade-off between energy consumption and performance. Additionally, we observed that workload awareness may significantly lower the overall number of migrations (5.47% – 37.58%), which has no adverse effect on application performance.