BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Admitting more tenants with tail latency SLOs  - Timothy Zhu\, CMU
DTSTART:20160425T123000Z
DTEND:20160425T133000Z
UID:TALK65894@talks.cam.ac.uk
CONTACT:44515
DESCRIPTION:Meeting tail latency Service Level Objectives (SLOs) in shared
  datacenter networks is known to be an important and challenging problem. 
 \nThe main challenge is in determining limits on the multi-tenancy such th
 at SLOs are met. This requires calculating latency guarantees\, which is a
  difficult problem\, especially when tenants exhibit bursty behavior as is
  common in production environments. Nevertheless\, recent papers in the pa
 st two years have shown techniques for calculating latency based on a bran
 ch of mathematical modeling called Deterministic Network Calculus (DNC). T
 he DNC theory is designed for adversarial worst-case conditions\, which is
  useful in some scenarios\, but is often overly conservative. Typical tena
 nts do not require strict worst-case guarantees\, but are only looking for
  SLOs at lower percentiles (e.g.\, 99th\, 99.9th). By considering SLOs at 
 lower percentiles\, it is possible to pack together many more tenants whil
 e still meeting tail latency SLOs. In this talk\, I'll present a brand new
  technique for calculating tail latency based on a probabilistic theory ca
 lled Stochastic Network Calculus (SNC). SNC is a new theory that is active
 ly being developed by the theory community to overcome the limitations of 
 DNC\, and we are the first to bring this theory to practice in a real comp
 uter system. In experiments on our cluster\, we demonstrate that our syste
 m can support twice as many tenants as the state-of-the-art while meeting 
 tail latency SLOs. \n
LOCATION:Auditorium\, Microsoft Research Ltd\, 21 Station Road\, Cambridge
 \, CB1 2FB
END:VEVENT
END:VCALENDAR