EON 2011 - EON 2011 : 2011 Workshop on Network-Optimized Computing at the Edge Of the Network

EON 2011

Network-Optimized Computing at the Edge Of the Network
Historically, there has been a distinct separation between the networking and server domains. This partitioning has reflected itself in networking processors being specifically optimized for layer 2/3/4 networking stacks up to TCP/IP and employing hardware acceleration to maintain line-rate processing. At the other end of the spectrum are general purpose server processors, which have a rich instruction-set architecture, programming and tools environment, but they require higher power and carry the overhead necessary for providing a general purpose processing infrastructure.

Network-optimized computing is the capability for application processing at networking speeds and unifies both networking and server functionalities. Network-optimized applications are constrained by ingress and egress data rates, latency and throughput requirements, and the temporal or streaming nature of the data. As the performance of network interconnects increases from current 10Gb/s rates and approaches 100Gb/s in 2015, network-optimizing requires a new generation of processor that can meet the latency and bandwidth needs while optimizing overall power dissipation. Important attributes of these converged systems include massive multithreading, integrated networking support, and hardware accelerators in a balanced system configuration with a robust instruction set architecture and programming environment.

Present and future network architectures consist of many layers, in which at each boundary or "edge," network-optimized computing can be carried out. This workshop will explore all aspects of processing at different edges in network architectures, each with it's own trade-offs and requirements.

Networks that are "wired" end-to-end are well-established, and the trend of edge-of-network computing has begun and growing in such networks, where the dominant method to optimize for end-to-end application performance is by utilizing a series of application optimization controllers at various edges of the network (e.g. XML acceleration, storage processing/acceleration in data-center and extended enterprise networks). Mobile computing, especially with higher data rates offered by current 3G and emerging 4G all-IP networks, offers the promise of seamlessly extending the wired experience to the wireless world. This workshop will especially encourage work that amplify and accelerate these trends. We also hope that the workshop will lead to the creation of an ecosystem of academic researchers and industrial practitioners that study related issues, share experiences, and document the trade-offs, for the benefit of the ISCA and extended high-performance computing community.

Date: June 4, 2011
Venue: San Jose Convention Center, San Jose, CA, USA.

This workshop will be held in conjunction with the International Symposium on Computer Architecture (ISCA) 2011, June 4-7, 2011 at the same location (San Jose Convention Center).