Abstract
The growth of cloud computing has resulted in uneconomic energy consumption, which has negatively impacted the environment through the generation of carbon emissions. This paper proposes a distributed Locust-inspired scheduling algorithm to reduce cloud computing consumed energy (LACE). LACE schedules and optimizes the allocation of virtual machines (VMs) based on behavior derived from locusts. LACE distributes scheduling among servers; each server is responsible for allocating and migrating its VMs. Hence, the scheduling load is distributed between servers rather than being centralized in one component. LACE was thoroughly evaluated by comparing it with long-standing VM scheduling algorithms: dynamic voltage-frequency scaling (DVFS), energy-aware scheduling using the workload-aware consolidation technique, and the static threshold with minimum utilization policy. The experimental results show that LACE considerably outperforms the other algorithms in almost every area. Most importantly, LACE exhibited significant levels of fault tolerance under heavy workloads that benchmarks were unable to sustain.