SIGCOMM '92 Conference proceedings on Communications architectures & protocols
IEEE/ACM Transactions on Networking (TON)
An engineering approach to computer networking: ATM networks, the Internet, and the telephone network
Scheduling Algorithms for Multiprogramming in a Hard-Real-Time Environment
Journal of the ACM (JACM)
Internetworking with TCP/IP, Volume 1: Principles, Protocols, and Architectures, Fourth Edition
Internetworking with TCP/IP, Volume 1: Principles, Protocols, and Architectures, Fourth Edition
Priority Inheritance Protocols: An Approach to Real-Time Synchronization
IEEE Transactions on Computers
High speed, scalable, and accurate implementation of packet fair queueing algorithms in ATM networks
ICNP '97 Proceedings of the 1997 International Conference on Network Protocols (ICNP '97)
An Efficient RMS Admission Control and its Application to Multiprocessor Scheduling
IPPS '98 Proceedings of the 12th. International Parallel Processing Symposium on International Parallel Processing Symposium
Hi-index | 14.98 |
This paper proposes a Versatile Network Architecture (${\cal V}$-net) to support flexible delay guarantees for applications in real-time networks. Applications communicate over the ${\cal V}$-net by using end-to-end network connections which support real-time and reliability characteristics tailored to meet the application's specified requirements. ${\cal V}$-net differs from other proposed architectures in that, in addition to addressing the issue of quality of service (QoS) feasibility for a wide spectrum of real-time applications, it also provides a mechanism to determine a network state dependent range of feasible delay values at each switching node along the routing path. These delay ranges can be used to assign per-node delays that reflect the resource availability of the node, thereby reducing the likelihood of bottlenecks along the routing path. The ${\cal V}$-net delay guarantees are provided for a variety of packet scheduling algorithms and traffic policing mechanisms. This flexibility is an important design consideration as a real-time network architecture must accommodate existing and future multimedia applications, with hard-, soft-, and non-real-time traffic. The performance evaluation results demonstrate the efficiency of this scheme in handling different traffic scenarios and QoS requirements. We have shown that it is possible, and indeed efficient, to determine an upper-bound on the delay of real-time traffic, when using our per-node delay assignment policies.