BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Analytics on Graphs with a Trillion Edges - Willy Zwaenepoel (EPFL
 )
DTSTART:20150508T100000Z
DTEND:20150508T110000Z
UID:TALK58093@talks.cam.ac.uk
CONTACT:Eiko Yoneki
DESCRIPTION:Big graphs occur naturally in many applications\, most obvious
 ly in social networks\, but also in many other areas such as biology and f
 orensics. Current approaches to processing large graphs use either superco
 mputers or very large clusters. In both cases the entire graph must reside
  in memory before it can be processed. We are pursuing an alternative appr
 oach\, processing graphs from secondary storage. While this comes with som
 e performance penalty\, it makes analytics on very large graphs feasible o
 n a small number of commodity machines. It also has the pleasing property 
 that "if you can store a graph\, you can compute on it". \n\nWe have devel
 oped two systems\, one for a single machine and one for a cluster of machi
 nes. X-Stream\, the single-machine solution\, aims to make all secondary s
 torage access sequential. It uses two techniques to achieve this goal: edg
 e-centric processing and streaming partitions. X-Stream outperforms the st
 ate-of-the-art GraphChi system\, because it achieves better sequentiality 
 and because it requires less preprocessing. Slipstream\, the cluster solut
 ion\, starts from the observation that there is little benefit to locality
  when accessing secondary storage over a high-speed network. As a result\,
  we use lightweight dynamic partitioning\, focusing on achieving load bala
 nce and sequential access to secondary storage. The resulting system achie
 ves good scaling and outperforms other systems. With Slipstream we have al
 so been able to process a trillion-edge graph\, a new milestone for graph 
 size on a small cluster. I will describe both systems and their performanc
 e on a number of benchmarks and in comparison to the state-of-the-art alte
 rnatives.\n\nThis work is joint work with Laurent Bindschaedler\, Jasmina 
 Malicevic and Amitabha Roy.\n\nBio: Willy Zwaenepoel received his BS/MS fr
 om the University of Gent\, Belgium\, and his PhD from Stanford University
 . He is currently a Professor of Computer Science at EPFL. Before he has h
 eld appointments as Professor of Computer Science and Electrical Engineeri
 ng at Rice University\, and as Dean of the School of Computer and Communic
 ation Sciences at EPFL. His interests are in operating systems and distrib
 uted systems. He is a Fellow of the ACM and the IEEE\, he has received the
  IEEE Kanai Award and several best paper awards\, and is a member of the B
 elgian and European Academies. He has also been involved in a number of st
 artups\, including iMimic (now part of Cisco)\, Midokura and Nutanix.\n
LOCATION:FW26\, Computer Laboratory\, William Gates Builiding
END:VEVENT
END:VCALENDAR