BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Latency-optimal fault-tolerant replication - Piotr Zielinski\, Inf
 erence Group
DTSTART:20060201T150000Z
DTEND:20060201T160000Z
UID:TALK4692@talks.cam.ac.uk
CONTACT:Phil Cowans
DESCRIPTION:In this talk\, I will present some of my PhD research in distr
 ibuted agreement protocols done at the Computer Laboratory\, Cambridge.\n\
 nSlides:\nhttp://www.cl.cam.ac.uk/~pz215/files/disc.pdf\nhttp://www.cl.cam
 .ac.uk/~pz215/files/msgeneric.pdf\n\nTo improve the fault-tolerance of a d
 istributed system\, a central\nserver can be replaced with many identical 
 replicas. Client requests\nare then broadcast to and executed by all these
 . To remain consistent\,\nall replicas must execute client requests in the
  same order.\n\nIn a simple approach\, clients send their requests to a di
 stinguished\nreplica\, which then broadcasts them in some order to all oth
 ers. This\nalgorithm is fast\; a client request reaches all replicas in on
 ly two\ncommunication steps. Unfortunately\, it is not fault tolerant\; a 
 crash\nof the distinguished replica will block the entire system.\nFault-t
 olerant solutions to this problem exist\, but they all require\nat least t
 hree steps\, even when no failures occur.\n\nIn this talk\, I will present
  new algorithms that combine the\nresistance of fault-tolerant solutions w
 ith the efficiency of\nfault-intolerant ones. Two facts are exploited: (i)
  requests from\ndifferent clients are usually independent\, and (ii) netwo
 rks normally\ndeliver messages to all recipients in the same order. In the
  absence\nof failures\, if one the above conditions holds\, these algorith
 ms\nrequire only two steps\, and three otherwise.\n\nKeywords: Atomic Broa
 dcast\, Generic Broadcast\, optimistic techniques\,\nreplication\, fault-t
 olerance\, Consensus\n
LOCATION:Committee Room\, Cavendish Laboratory
END:VEVENT
END:VCALENDAR