GouicemRedha, CarverDamien, LoziJean-Pierre, SopenaJulien, LepersBaptiste, ZwaenepoelWilly, PalixNicolas, LawallJulia, MullerGilles (2020), Fewer Cores, More Hertz: Leveraging High-Fequency Cores in the OS Scheduler for Improved Application Performance, in Proceedings ATC 2020
, Usenix ATC 2020, USA.
Lepers Baptiste, Gouicem Redha, Carver Damien, Lozi Jean-Pierre, Palix Nicolas, Apont Maria-Virginia, Zwaenepoel Willy, Sopena Julien, Lawall Julia, Muller Gilles (2020), Provable Multicore Schedulers with Ipanema: Application to Work Conservation, in EuroSys 2020
, EuroSys 2020, Heraklion, Crete, Greece.
Bindschaedler Laurent, Goel Ashvin, Zwaenepoel Willy (2020), Hailstorm: Disaggregated Compute and Storage for Distributed LSM-based Databases, in ASPLOS 2020
, ASPLOS 2020 25th ACM International Conference on Architectural Support for Programming Languages and, Lausanne, Switzerland.
LepersBaptiste, BalmauOana, GuptaKaran, ZwaenepoelWilly (2019), Kvell: The Design and Implementation of a Fast Persistent Key-Value Store, in Proceedings SOSP
, SOSP 2019 - The 27th ACM Symposium on Operating Systems Principles, Huntsville, Ontario, Canada.
BouronJustinien, ChevalleySébastien, LepersBaptiste, ZwaenepoelWilly (2018), The Battle of the Schedulers: FreeBSD ULE vs Linux CFS, in Proceedings Usenix 2018
Laurent Bindschaedler Jasmina Malicevic Nicolas Schiper Ashvin Goel and Willy Zwaenepoel (2018), Rock You like a Hurricane: Taming Skew in Large Scale Analytics, in Eurosys 2018
, Eurosys 2018, Eurosys 2018.
Jasmina Malicevic Baptiste Lepers and Willy Zwaenepoel (2017), Everything you always wanted to know about multicore graph processing but were afraid to ask, in USENIX ATC'17
, Conference USENIX ATC'17, Santa Clara, California, USA.
Analytics over large graphs is attracting increasing attention. Part of the reason for this upsurge of interest is the great variety of information that is naturally encoded as graphs. Large graphs are obviously present in social networks, but they also occur naturally in many other applications, for example, in biology, forensics, or logistics. Graph processing poses an interesting systems challenge: graph algorithms tend to exhibit little locality, making it difficult to build platforms that exhibit good performance. Much progress has been made in recent years in building such systems on platforms ranging from supercomputers over large clusters to single (multicore) machines, using either in-memory or out-of-core approaches. Many of these first-generation systems are, however, rather inflexible, restricting users to a particular environment and computation on static graphs.Based on our earlier work on out-of-core graph processing systems, we propose to advance the state of the art in graph processing by building systems that gracefully scale between memory and storage and that are capable of dealing with dynamic graphs. In addition, we intend to further optimize out-of-core performance, both in terms of performance and capacity.