Energy consumption of scheduling policies for HTC jobs in the cloud

Abstract

Organisations often exploit idle time on existing local computing infrastructure through High Throughput Computing (HTC) to perform computation. More recently the same approach has been employed to make use of cloud resources in large-scale computation. To date, the impact of HTC scheduling policies within a cloud environment have received limited attention in the literature. The key focus of this paper is to extend an existing cloud simulation system to incorporate energy measurement, and evaluate the energy and performance impact of existing policies for scheduling HTC jobs to cloud resources. We demonstrate through trace-driven simulation the trade-off between energy consumption and system performance for a number of HTC scheduling policies.