Dynamic global resource allocation in shared data centers and clouds

  • Authors:

  • Gokul Soundararajan;Saeed Ghanbari;Daniel Lupei;Jin Chen;Cristiana Amza

  • Affiliations:

  • University of Toronto, IBM Canada, CAS Research, Markham, Ontario, Canada;University of Toronto, IBM Canada, CAS Research, Markham, Ontario, Canada;University of Toronto, IBM Canada, CAS Research, Markham, Ontario, Canada;University of Toronto, IBM Canada, CAS Research, Markham, Ontario, Canada;University of Toronto, IBM Canada, CAS Research, Markham, Ontario, Canada

  • Venue:
  • CASCON '12 Proceedings of the 2012 Conference of the Center for Advanced Studies on Collaborative Research
  • Year:
  • 2012

Quantified Score

Hi-index 0.00

Visualization

Abstract

We design, implement and evaluate a global resource allocator to provide end-to-end quality of service in shared data centers and Clouds. Global resource allocation involves performance modeling for proportioning several levels of storage cache, and the storage bandwidth between applications according to overall performance goals. The problem is challenging due to the interplay between different resources, e.g., changing the client cache quota affects the access pattern at the cache/disk levels below it in the storage hierarchy. We use a combination of on-line modeling and sampling to arrive at near-optimal configurations within minutes. The key idea is to incorporate access tracking and known resource dependencies e.g., due to cache replacement policies, into our performance model. In our experimental evaluation, we use both micro-benchmarks and the industry standard benchmarks TPC-W and RUBiS. We show that our global resource allocation approach provides up to a factor of 1.4 better overall performance compared to a combination of state-of-the-art single resource controllers, at the same cost.