Performance evaluation of scheduling algorithms for database services with soft and hard SLAs

  • Authors:
  • Hyun Jin Moon;Yun Chi;Hakan Hacigümüş

  • Affiliations:
  • NEC Laboratories America, Cupertino, CA, USA;NEC Laboratories America, Cupertino, CA, USA;NEC Laboratories America, Cupertino, CA, USA

  • Venue:
  • Proceedings of the second international workshop on Data intensive computing in the clouds
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

Database service providers face heterogeneous customer workloads with widely varying characteristics. The query scheduling plays a critical role in serving such workloads and involves the careful consideration of specific requirements introduced in service provisioning environments. As a part of our real platform building process, we have collected specific requirements for the scheduling framework through extensive interactions with business organizations that provide services to real clients. Although there is a very large body of previous work in the scheduling area, there is no single scheduling method that is designed to satisfy all of our requirements. However, some of them may address certain aspects of those requirements. In this work, we rigorously evaluate a comprehensive set of such scheduling methods and present how they perform with respect to the full requirements list. We also propose and evaluate an effective extension to the most promising method, iCBS, we identified through the evaluations in this space.