BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:HovercRaft: Achieving Scalability and Fault-tolerance for microsec
 ond-scale Datacenter Services - Marios Kogias
DTSTART:20200507T140000Z
DTEND:20200507T150000Z
UID:TALK142120@talks.cam.ac.uk
CONTACT:Srinivasan Keshav
DESCRIPTION:Cloud platform services must simultaneously be scalable\, meet
  low tail latency service-level objectives\, and be resilient to a combina
 tion of software\, hardware\, and network failures. Replication plays a fu
 ndamental role in meeting both the scalability and the fault-tolerance req
 uirement\, but is subject to opposing requirements: (1) scalability is typ
 ically achieved by relaxing consistency\; (2) fault-tolerance is typically
  achieved through the consistent replication of state machines. Adding nod
 es to a system can therefore either increase performance at the expense of
  consistency\, or increase resiliency at the expense of performance.\nWe p
 ropose HovercRaft\, a new approach by which adding nodes increases both th
 e resilience and the performance of general-purpose state-machine replicat
 ion. We achieve this through an extension of the Raft protocol that carefu
 lly eliminates CPU and I/O bottlenecks and load balances requests.\nOur im
 plementation uses state-of-the-art kernel-bypass techniques\, datacenter t
 ransport protocols\, and in-network programmability to deliver up to 1 mil
 lion operations/second for clusters of up to 9 nodes\, linear speedup over
  unreplicated configuration for selected workloads\, and a 4X speedup for 
 the YCSBE-E benchmark running on Redis over an unreplicated deployment.\n\
 nhttps://meetingsemea10.webex.com/meetingsemea10/j.php?MTID=mf0c350ff1fab9
 6540218f5767c51d098
LOCATION:Webex Meeting number: 140 388 203 Password: ZApmfTb7h56
END:VEVENT
END:VCALENDAR