Thermal Modeling of Hybrid Storage Clusters

  • Authors:

  • Xunfei Jiang;Maen M. Al Assaf;Ji Zhang;Mohammed I. Alghamdi;Xiaojun Ruan;Tausif Muzaffar;Xiao Qin

  • Affiliations:

  • Department of Computer Science and Software Engineering, Auburn University, Auburn, USA 36849-5347;King Abdullah II School for Information Technology, The University of Jordan, Amman, Jordan;Department of Computer Science and Software Engineering, Auburn University, Auburn, USA 36849-5347;Department of Computer Science, Al-Baha University, Al-Baha City, Kingdom of Saudi Arabia;Department of Computer Science, West Chester University of Pennsylvania, West Chester, USA 19383;Department of Computer Science and Software Engineering, Auburn University, Auburn, USA 36849-5347;Department of Computer Science and Software Engineering, Auburn University, Auburn, USA 36849-5347

  • Venue:
  • Journal of Signal Processing Systems
  • Year:
  • 2013

Quantified Score

Hi-index 0.00

Visualization

Abstract

There is a lack of thermal models for storage clusters; most existing thermal models do not take into account the utilization of hard drives (HDDs) and solid state disks (SSDs). To address this problem, we build a thermal model for hybrid storage clusters that are comprised of HDDs and SSDs. We start this study by generating the thermal profiles of hard drives and solid state disks. The profiling results show that both HDDs and SSDs have profound impacts on temperatures of storage nodes in a cluster. Next, we build two types of hybrid storage clusters, namely, inter-node and intra-node hybrid storage clusters. We develop a model to estimate the cooling cost of a storage cluster equipped with hybrid storage nodes. The thermal model is validated against data acquired by temperature sensors. Experimental results show that, compared to the HDD-first strategy, the SSD-first strategy is an efficient approach to minimize negative thermal impacts of hybrid storage clusters.