On Maximum Rate Control of Weighted Fair Scheduling for Transactional Systems

  • Authors:
  • Jeng Farn Lee;Yeali Sun;Meng Chang Chen

  • Affiliations:
  • -;-;-

  • Venue:
  • RTSS '03 Proceedings of the 24th IEEE International Real-Time Systems Symposium
  • Year:
  • 2003

Quantified Score

Hi-index 0.00

Visualization

Abstract

While exisiting weighted fair scheduling schemes guaranteeminimum service rates of a shared server (such as a computeror a communication channel), maximum rate control wasgenerally enforced by employing policing mechanisms. Theprevious approaches use either a concatenation of ratecontroller and scheduler, or a policer in front of scheduler.The concatenation method uses two sets of queues andmanagement aparartus, and thus incurs overhead. The othermethod allows bursty job requests that may violatemaximum rate constraint. In this paper, we present a newweighted fair scheduling scheme, WF2Q-M, to simultaneouslysupport maximum rate control and minimum service rateguarantee. WF2Q-M proposes the virtual clock adjustmentmethod to enforce maximum rate control by distributing theexcess bandwidths of maximum rate constrained sessions toother sessions without recalculating the virtual starting andfinishing times of regular sessions. In terms of performancemetrics, we prove that WF2Q-M is theoretically bounded bya fluid reference model. A procedural schedulingimplementation of WF2Q-M is proposed and proof ofcorrectness is given. Finally, we conduct extensiveexperiments to show the performance of WF2Q-M is just as weclaim.