An energy-efficiency-aware resource allocation strategy in multi-granularity provision for green computing

Energy efficiency is one of the most important issues in current large-scale server systems for green computing. A reasonable resource allocation is a key factor impacting on the energy efficiency of a system. Existing scheduling mechanisms in Map Reduce environments focus on the fair sharing of cluster resources among multiple users, allocated resources and scheduled jobs by a priority-based strategy. However, such mechanisms have little awareness of users' SLA (Service-Level Agreement) . It is difficult to map the user's SLA to a certain priority. Additionally, they can't be sensitive to the energy-efficiency changes while a cluster is running and a work is going. Also users' SLA cannot be satisfied accurately and effectively. In this paper, an energyefficiency-aware scheduling mechanism of a virtual resource for a big data application is addressed. This mechanism calculates the energy-efficiency value of different granularity resource mappings. Based on the proposed strategy, we also designed an optimized resource allocation algorithm for different service requirements, which can achieve the maximal energy-efficiency value for green computing.