BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Chicago
X-LIC-LOCATION:America/Chicago
BEGIN:DAYLIGHT
TZOFFSETFROM:-0600
TZOFFSETTO:-0500
TZNAME:CDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0500
TZOFFSETTO:-0600
TZNAME:CST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260202T201247Z
LOCATION:Second Floor Atrium
DTSTART;TZID=America/Chicago:20251120T080000
DTEND;TZID=America/Chicago:20251120T170000
UID:submissions.supercomputing.org_SC25_sess533_post232@linklings.com
SUMMARY:CATIOS: Time-Resolved I/O-Aware Job Scheduling for HPC Systems
DESCRIPTION:YuTsen Tseng (Tohoku University, Graduate School of Informatio
 n Sciences); Masatoshi Kawai (Tohoku University); Keichi Takahashi (Univer
 sity of Osaka); and Hiroyuki Takizawa (Tohoku University)\n\nHPC workloads
  are increasingly data-intensive, with contention on shared storage emergi
 ng as a primary bottleneck. Existing I/O-aware job schedulers rely on stat
 ic bandwidth assumptions that overlook time-varying I/O behavior, leading 
 to inefficient utilization and unpredictability.\n\nThis work introduces t
 he Contention-Avoiding Temporal I/O-aware job Scheduling (CATIOS) framewor
 k, which considers temporal I/O behavior in scheduling decisions to addres
 s these issues. CATIOS first matches incoming jobs with similar historical
  profiles semantically and resource-wise, who serve as proxies and are eva
 luated sequentially in time-resolved contention-aware simulation after bei
 ng reprioritized by configurable scheduling policies. These enable CATIOS 
 to avoid overlapping bursts according to objectives.\n\nEvaluations with B
 lue Waters workloads on a SimGrid-based platform show that CATIOS reduces 
 makespan while maintaining controlled average wait times, achieving a bala
 nced trade-off of approximately 1:1.3 between makespan reduction and avera
 ge wait time increase. These results demonstrate CATIOS’s capability to im
 prove data-intensive HPC systems with mixed workloads.\n\nTag: Research & 
 ACM SRC Posters\n\nRegistration Category: Technical Program Reg Pass\n\nSe
 ssion Chairs: Kento Sato (RIKEN Center for Computational Science (R-CCS));
  Chris Schlipalius (Pawsey Supercomputing Research Centre; Commonwealth Sc
 ientific and Industrial Research Organisation (CSIRO), Australia); and Anj
 a Gerbes (Georg-August-Universität Göttingen)\n\n
END:VEVENT
END:VCALENDAR