The Tempest: a framework for safe, resource assured, programmable networks
IEEE Communications Magazine
Architectural analysis and instruction-set optimization for design of network protocol processors
Proceedings of the 1st IEEE/ACM/IFIP international conference on Hardware/software codesign and system synthesis
Balancing performance and flexibility with hardware support for network architectures
ACM Transactions on Computer Systems (TOCS)
Design of an efficient memory subsystem for network processor
Proceedings of the 2005 Asia and South Pacific Design Automation Conference
Hi-index | 4.10 |
The convergence of computing and networking is nowhere more evident than in the phenomenal growth of the World Wide Web. In another sense, though, computer networking is being pulled in two opposite directions. On the one hand, the Web's popularity and growth has been fueled largely by desk-top applications consuming bandwidth-intensive images and video. On the other hand, thin-client computers are becoming more commonly used as edge-of-network devices, often connected by wireless technology. There is also an increasing mismatch between fiber-optic transmission bandwidths and computer speeds, pushing computing further away from the network core. Are there ways to close, or at least manage, this growing schism-whether through novel hardware solutions or the programmability of network infrastructures? Can we better integrate these edge-of- network devices and make them full-fledged network participants? High-speed optical networks mix many types of traffic and services, so only features needed by all aggregated services will be cost-effective to place there. They are unlikely to employ software and computing except in the so-called control plane, which manipulates the behavior of the transport system. Low-cost and low-powered thin clients need networks and software that allow plug and play and unplug: secure, mobile, sporadically detached operation. Introducing programmability into the network infrastructure (or exposing preexisting programmability to users) could meet these needs, and active networks are one model for programming net-works. Although what technologies will prevail is in question, networking and computing will continue to converge. Currently, this convergence is resulting in increased collaboration between the network infrastructure and the end hosts to meet application needs.