BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Caching with Delayed Hits - Nirav Atre\, CMU
DTSTART:20200730T140000Z
DTEND:20200730T150000Z
UID:TALK149731@talks.cam.ac.uk
CONTACT:Srinivasan Keshav
DESCRIPTION:Caches are at the heart of latency-sensitive systems. In this 
 talk\, we will focus on a growing challenge for the design of latency-mini
 mizing caching called ‘delayed hits’. Delayed hits occur at high throu
 ghput\, when multiple requests to the same object queue up before an outst
 anding cache miss for that object is resolved. This phenomenon increases l
 atencies beyond the predictions of traditional caching models and simulato
 rs\, and subverts expectations of existing caching algorithms\; in fact\, 
 caching algorithms are designed as if delayed hits simply didn't exist. We
  show that traditional caching strategies – even so-called ‘optimal’
  algorithms – can fail to minimize latency in the presence of delayed hi
 ts. We present a new\, latency-optimal offline caching algorithm called BE
 LATEDLY\, which computes up to 45% lower latencies compared to the traditi
 onal\, hit-rate optimal Belady’s algorithm. Using BELATEDLY as our guide
 \, we show that incorporating an object’s ‘Aggregate Delay’ into onl
 ine caching heuristics can improve latencies for practical caching systems
  by up to 40%. We implement a prototype\, Minimum-AggregateDelay (MAD)\, i
 n the context of a CDN caching node. Using a CDN production trace and back
 ends deployed in different geographic locations\, we show that MAD can red
 uce average caching latencies in CDNs today by 12-18% depending on the bac
 kend RTTs.\n\nBio: Nirav is a third-year Ph.D. student in the Computer Sci
 ence Department (CSD) at Carnegie Mellon University (CMU)\, where he is ad
 vised by Prof. Justine Sherry. Nirav's research interests lie at the inter
 section of networking and performance modeling\, and he's part of the Syst
 ems\, Networking\, and Performance (SNAP) Lab at CMU. Prior to starting gr
 aduate school\, Nirav completed his B.A.Sc in Computer Engineering at the 
 University of Toronto\, Canada\, in 2018.
LOCATION:https://meet.google.com/ehj-dwaz-rea
END:VEVENT
END:VCALENDAR
