Efficient Management of System Idleness
Ningfang Mi, Alma Riska, Qi Zhang, Evgenia Smirni, Erik Riedel · 2008
AbstractVarious activities that intend to enhance performance, reliability, and availability of systems are scheduled withlow priority and served during system idle times. Under such conditions, idleness becomes a valuable “resource”that needs to be efficiently managed. A common approach in sys tem design is to be non-work-conserving by “idlewaiting”, i.e., delay scheduling of background jobs to avoid slowing down upcoming foreground tasks.In this paper, we complement “idle-waiting” with the “estimation” of background work to be served in every idleinterval to manage the trade-off between the performance offoreground and background tasks. As a result, the overallsystem is better utilized without compromising foreground performance. Our analysis shows that if idle times havelow variability, then it is not necessary to idle wait before starting a background job. Only if idle times are highlyvariable, then idle waiting is necessary to minimize the impact of background activity on foreground performance.We further show that if there is burstiness in addition to high variability in idle intervals, then it is possible to predictaccurately the length of incoming idle times and use that information to serve more background jobs without affectingforeground performance.