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:20211207T055347Z
LOCATION:Online
DTSTART;TZID=America/Chicago:20211115T113000
DTEND;TZID=America/Chicago:20211115T120000
UID:submissions.supercomputing.org_SC21_sess422_ws_pmbsf107@linklings.com
SUMMARY:Exploration of Congestion Control Techniques on Dragonfly-Class HP
 C Networks Through Simulation
DESCRIPTION:Workshop\n\nExploration of Congestion Control Techniques on Dr
 agonfly-Class HPC Networks Through Simulation\n\nMcGlohon, Hemmert, Brown,
  Levenhagen, Chunduri...\n\nEnsuring optimal communication latency in high
 -performance computing (HPC) networks is of critical importance to the eff
 icient operation of facilitated applications. Different application operat
 ions and types of tasks, such as I/O operations, can create a variety of t
 raffic patterns across the system interconnect. Some communication pattern
 s, however, can be problematic for overall system performance.\n\nOne traf
 fic pattern of particular concern is the many-to-one or incast. When packe
 ts sent from many different endpoints target a singular destination, or a 
 small number of destinations, they can overwhelm the receiving endpoints' 
 ability to process the traffic, resulting in a cascading effect of induced
  congestion. This can have broad-reaching, detrimental effects to other ap
 plications as their data streams encounter induced congestion.\n\nThe conc
 ept of congestion control has been explored in various HPC system technolo
 gies and is an important feature in state-of-the-art networks such as Infi
 niband and the Cray Slingshot interconnect. Because access to physical, fu
 ll-scale interconnects of bleeding-edge design can be challenging and the 
 exact mechanisms of operation not publicly known, we look to simulation to
  explore techniques for congestion control with a fine level of flexibilit
 y not available on real-world systems.\n\nWe present and explore a mechani
 sm for congestion control which seeks to detect network congestion, identi
 fy its cause and abate it by throttling injection of identified aggressor 
 endpoints. Our work proposes, discusses and evaluates two similar implemen
 tations of this mechanism for congestion control in two different network 
 simulators and their ability to mitigate the effects of congestion on appl
 ication communication performance and general system packet latencies.\n\n
 Tag: Online Only, Accelerator-based Architectures, Applications, Computati
 onal Science, Emerging Technologies, Extreme Scale Computing, File Systems
  and I/O, Heterogeneous Systems, Parallel Programming Languages and Models
 , Performance, Scientific Computing, Software Engineering\n\nRegistration 
 Category: Workshop Reg Pass
END:VEVENT
END:VCALENDAR