Title: Exploiting Fine-Grain Idle Time on a Cluster of Workstations

Speaker: Kyung Dong Ryu

Abstract:
Studies have shown that a significant fraction of the time, workstations are
idle. Traditional idle resource harvesting systems define a social contract
that permits foreign jobs to run only when a workstation is idle. In this
talk, I will first present a new policy, called Linger-Longer, that refines
the social contract to permit fine grained cycle stealing. Linger-Longer
allows foreign jobs to linger on a machine at low priority even when local
tasks are active. Our simulation study shows that the Linger-Longer policy
can improve the throughput of foreign jobs on a cluster by up to 60% with
only a few percent slowdown of local jobs. The simulation also demonstrates
that guest parallel jobs can perform better with our new approach than with
the traditional run-time reconfiguration approach. I will also present a set
of mechanisms (implemented as Linux kernel extensions) that allow the
operating systems to limit guest jobs' consumption of resources such as CPU
cycles, physical memory, and I/O and network bandwidth.