BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IC Monday Seminars : Robust Replication (or how I learned to stop 
 worrying and love failures)
DTSTART:20120326T161500
DTSTAMP:20260407T065950Z
UID:206906a5f360dce13988ad0ce501ffae1be9ea9219bcbebf58052e1b
CATEGORIES:Conferences - Seminars
DESCRIPTION:Dr. Allen Clement\, Max Planck Institute for Software Systems 
 - IC Faculty candidate\nAbstract\nThe choice between Byzantine and crash f
 ault tolerance is viewed as a fundamental design decision when building fa
 ult tolerant systems. We show that this dichotomy is not fundamental\, and
  present a unified model of fault tolerance in which the number of tolerat
 ed faults of each type is a configuration choice.  Additionally\, we obse
 rve that a single fault is capable of devastating the performance of exist
 ing Byzantine fault tolerant replication systems.  We argue that fault to
 lerant systems should\, and can\, be designed to perform well even when fa
 ilures occur.  In this talk I will expand on these two insights and descr
 ibe our experience leveraging them to build a generic fault tolerant repli
 cation library that provides flexible fault tolerance and robust performan
 ce.  We use the library to build a (Byzantine) fault tolerant version of 
 the Hadoop Distributed File System.\n\nBiography\nAllen Clement is a Post-
 doctoral Research Fellow at the Max Planck Institute for Software Systems.
  He received a Ph.D. from the University of Texas at Austin in 2010 and a
 n A.B. in Computer Science from Princeton University.  His research focus
 es on the challenges of building robust and reliable distributed systems.
   In particular\, he has investigated practical Byzantine fault tolerant 
 replication\, systems robust to both Byzantine and selfish behaviors\, con
 sistency in geo-replicated environments\, and how to leverage the structur
 e of social networks to build Sybil-tolerant systems.
LOCATION:INM 202
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR