BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:Europe/Stockholm
X-LIC-LOCATION:Europe/Stockholm
BEGIN:DAYLIGHT
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260625T133339Z
LOCATION:Bldg. 6 - 001 - Plenary Room
DTSTART;TZID=Europe/Stockholm:20260630T124000
DTEND;TZID=Europe/Stockholm:20260630T124100
UID:submissions.pasc-conference.org_PASC26_sess129_posC103@linklings.com
SUMMARY:ACMP11 - Mosaic: Automatic Categorization of I/O Patterns from Sci
 entific Applications
DESCRIPTION:Théo Jolivel (INRIA, Universite de Rennes)\n\nWhile newly depl
 oyed High-Performance Computing (HPC) systems embeds more compute capabili
 ties than the generations before, parallel file systems (PFS) are struggli
 ng to keep up with this same trend, increasing the gap between computing p
 ower and I/O performance. If new paradigms and technologies are implemente
 d to provide faster storage accesses, like tier-based systems and local st
 orage buffers, it is essential to understand how the I/O behavior of appli
 cations to take advantage of these resources. Such knowledge facilitates t
 he design of efficient I/O optimizations, and enables to better understand
  the loads experienced by the PFS.\n  In this context, we developed MOSAIC
 , a Python library that processes I/O traces collected at machine scale an
 d classify them based on I/O patterns. MOSAIC extracts patterns through th
 ree pillars: I/O temporality, I/O periodicity, and metadata requests. This
  allows a high-level understanding of how and when applications need to ac
 cess data and the evolution of those patterns across time and machines.\n 
  We validate MOSAIC on datasets from two top-tier supercomputers, unveilin
 g a clear correlation between large data writes and periodicity. We also e
 xplore pattern variations across multiple runs of the same application or 
 launched by the same user.\n\nSession Chair: Tobias Hodel (University of B
 ern, Switzerland)\n\n
END:VEVENT
END:VCALENDAR