BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20200129T163557Z
LOCATION:405-406-407
DTSTART;TZID=America/Denver:20191122T112900
DTEND;TZID=America/Denver:20191122T114400
UID:submissions.supercomputing.org_SC19_sess129_ws_hpcsysp112@linklings.co
 m
SUMMARY:A Better Way of Scheduling Jobs on HPC Systems: Simultaneous Fair-
 Share
DESCRIPTION:Workshop\n\nA Better Way of Scheduling Jobs on HPC Systems: Si
 multaneous Fair-Share\n\nSteffen\n\nTypical HPC job scheduler software det
 ermines scheduling order by a linear sum of weighted priority terms. When 
 a system has a rich mix of job types, this makes it difficult to maintain 
 good productivity across diverse user groups. Currently-implemented fair-s
 hare algorithms tweak priority calculations based on \emph{past} job handl
 ing by modifying priority, but don't fully solve problems of queue-stuffin
 g and various classes of under-served job types, because of coupling betwe
 en different terms of the linear calculated priority . <br /> <br />This p
 aper proposes a new scheme of scheduling jobs on an HPC system called "Sim
 ultaneous Fair-share'' (or "SFS'') that works by considering the jobs alre
 ady committed to run in a given time slice and adjusting which jobs are se
 lected to run accordingly. This allows high-throughput collaborations to g
 et lots of jobs run, but avoids the problems of some groups starving out o
 thers due to job characteristics, all while keeping system administrators 
 from having to directly manage job schedules. This paper presents Simultan
 eous Fair-share in detail, with examples, and shows testing results using 
 a job throughput and scheduler simulation.\n\nTag: Workshop Reg Pass, Data
 center, SIGHPC Workshop, State of the Practice, System Administration, Sys
 tem Maintenance, System Reliability\n\nRegistration Category: Workshop Reg
  Pass, Datacenter, SIGHPC Workshop, State of the Practice, System Administ
 ration, System Maintenance, System Reliability
URL:https://sc19.supercomputing.org/presentation/?id=ws_hpcsysp112&sess=se
 ss129
END:VEVENT
END:VCALENDAR