BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Naiad: A Timely Dataflow System - Derek Murray (MSR Silicon Valley
 )
DTSTART:20130925T133000Z
DTEND:20130925T143000Z
UID:TALK46908@talks.cam.ac.uk
CONTACT:Eiko Yoneki
DESCRIPTION:Naiad is a distributed system for executing data parallel\,\nc
 yclic dataflow programs. It offers the high throughput\nof batch processor
 s\, the low latency of stream processors\,\nand the ability to perform ite
 rative and incremental\ncomputations. Although existing systems offer some
  of\nthese features\, applications that require all three have relied\non 
 multiple platforms\, at the expense of efficiency\,\nmaintainability\, and
  simplicity. Naiad resolves the complexities\nof combining these features 
 in one framework.\nA new computational model\, timely dataflow\, underlies
 \nNaiad and captures opportunities for parallelism\nacross a wide class of
  algorithms. This model enriches\ndataflow computation with timestamps tha
 t represent\nlogical points in the computation and provide the basis\nfor 
 an efficient\, lightweight coordination mechanism.\nWe show that many powe
 rful high-level programming\nmodels can be built on Naiad’s low-level pr
 imitives\, enabling\nsuch diverse tasks as streaming data analysis\, itera
 tive\nmachine learning\, and interactive graph mining.\nNaiad outperforms 
 specialized systems in their target application\ndomains\, and its unique 
 features enable the development\nof new high-performance applications.\n\n
 Bio: Derek Murray is on the cusp of becoming a full-time researcher at the
  Silicon Valley Lab of Microsoft Research. During his postdoc there\, he w
 orked mainly on the Naiad project\, which he will be presenting at SOSP 20
 13. He completed his PhD at the University of Cambridge in 2011\, under th
 e supervision of Steve Hand.\n\n
LOCATION:LT2\, Computer Laboratory\, William Gates Builiding
END:VEVENT
END:VCALENDAR