Core-stateless Guaranteed Services Networks
A Core-stateless Guaranteed Services Network Architecture
We have proposed link scheduling mechanisms that enable networks to provide performance guarantees on a per-flow basis, without maintaining or using any per-flow state in the core routers. We have conducted extensive theoretical analysis to establish the end-to-end properties exported by the resultant Core-stateless Guaranteed Services (CSGS) network architecture. We are currently implementing and evaluating the performance and scalability of routers in a CSGS network using a programmable router platform.
A complete functional network requires other components, in addition to those we have proposed. For instance, a CSGS network provides upper bounds on end-to-end delay and throughput only under the assumption that the sum of reserved rates does not exceed the capacity of any link. Thus, admission control is an essential part of a CSGS network. To efficiently implement such additional components, we leverage existing research on efficient admission control, packet sorting, buffer management, end-to-end signaling, and packet classification.
We have completed a base implementation and preliminary evaluation of the CSGS architecture using Intel's IXP1200-based router platform. We choose this platform because it offers an open architecture with programmability both in the control plane as well as the date path. One of the main implementation challenges is to enable CSGS routers to operate at high link speeds. Our performance tests indicate that core routers in a CSGS network can process packets at 75% of the speeds of conventional IP routers under all kinds of workload conditions, and under more favourable conditions, can come within 10% of the throughput performance of current IP routers.
Relevant Publications :
- J. Kaur and H. Vin, "Providing Deterministic End-to-end Fairness Guarantees in Core-stateless Networks" , in Proceedings of the Eleventh International Workshop on Quality of Service (IWQoS'03), Monterey, CA, June 2003.
- J. Kaur and H. Vin, "Core-stateless Guaranteed Throughput Networks" , in Proceedings of IEEE INFOCOM, San Francisco, CA, April 2003.
- J. Kaur and H. Vin, "End-to-end Fairness Analysis of Fair Queuing Networks" , in the 23rd IEEE International Real-Time Systems Symposium (RTSS'02), Austin, TX, Dec 2002.
- J. Kaur, "Scalable Network Architectures for Providing Per-flow Service Guarantees" , Ph.D. Dissertation, Department of Computer Sciences, University of Texas at Austin, Aug 2002.
- J. Kaur and H. Vin, "Core-Stateless Guaranteed Rate Scheduling Algorithms" , in Proceedings of IEEE INFOCOM, Anchorage, AK, April 2001.
- V. Sundaram, A. Chandra, P. Goyal, P. Shenoy, J. Sahni, and H. Vin, "Application Performance in the QLinux Multimedia Operating System" , in Proceedings of the Eighth ACM Conference on Multimedia, Los Angeles, CA, November 2000.
- J. Sahni, P. Goyal, and H. Vin, "Scheduling CBR Flows: FIFO or Per-Flow Queueing?" , in Proceedings of the Ninth IEEE International Workshop on Network and Operating System Support for Digital Audio and Video (NOSSDAV'99), Basking Ridge, NJ, June 1999.