EPOBF: Energy Efficient Allocation of Virtual Machines in High Performance Computing Cloud

Abstract

Cloud computing has become more popular in provision of computing resources under virtual machine (VM) abstraction for high performance computing (HPC) users to run their applications. A HPC cloud is such cloud computing environment. One of challenges of energyefficient resource allocation for VMs in HPC cloud is trade-off between minimizing total energy consumption of physical machines (PMs) and satisfying Quality of Service (e.g. performance). On one hand, cloud providers want to maximize their profit by reducing the power cost (e.g. using the smallest number of running PMs). On the other hand, cloud customers (users) want highest performance for their applications. In this paper, we focus on the scenario that scheduler does not know global information about user jobs and/or user applications in the future. Users will request short-term resources at fixed start-times and non-interrupted durations. We then propose a new allocation heuristic (named Energy-aware and Performance-per-watt oriented Best-fit (EPOBF)) that uses metric of performance-per-watt to choose which most energyefficient PM for mapping each VM (e.g. maximum of MIPS/Watt). Using information from Feitelson's Parallel Workload Archive to model HPC jobs, we compare the proposed EPOBF to state-of-the-art heuristics on heterogeneous PMs (each PM has multicore CPU). Simulations show that the EPOBF can reduce significant total energy consumption in comparison with stateof-the-art allocation heuristics.

A GPU-Based Enhanced Genetic Algorithm for Power-Aware Task Scheduling Problem in HPC Cloud

“…In the future work, we will concern on some real constraints (as in [11]) on the PATS and we will investigate on improving quality of chromosomes (solutions) by applying EPOBF heuristic in [12] and Memetic methodology in each genetic operation.…”

Energy efficiency of VM consolidation in IaaS clouds

“…Verma et al [35] applied First Fit (FF) descending to generate an initial assignment of VMs and later used Best Fit (BF) to reassign VMs picked from servers that violate resource constraints. Goiri et al [11] and Quang-Hung et al [28] studied corebased heuristic and BF descending heuristic to choose the most energy-efficient server for mapping each VM, respectively. These heuristic methods are applicable to stateless requests [34], based on the recognition that we can use the capacities of a bin to the fullest extent after the current allocation is performed.…”

On Achieving Energy Efficiency and Reducing CO2Footprint in Cloud Computing

“…One such approach [7] evaluates the performance-per-watt achieved by a range of VM allocation heuristics and shows that the best available algorithms offer substantial reductions in energy usage. In common with many such approaches, these results do not translate energy consumption into environmental impact, were obtained from simulation rather than experimentation and do not address the need for post deployment management of applications.…”