TCP performance in ATM networks: ABR parameter tuning and ABR/UBR comparisons
- Sandia National Labs., Livermore, CA (United States)
- Cisco Systems, Inc., San Jose, CA (United States)
This paper explores two issues on TOP performance over ATM networks: ABR parameter tuning and performance comparison of binary mode ABR with enhanced UBR services. Of the fifteen parameters defined for ABR, two parameters dominate binary mode ABR performance: Rate Increase Factor (RIF) and Rate Decrease Factor (RDF). Using simulations, we study the effects of these two parameters on TOP over ABR performance. We compare TOP performance with different ABR parameter settings in terms of through-puts and fairness. The effects of different buffer sizes and LAN/WAN distances are also examined. We then compare TOP performance with the best ABR parameter setting with corresponding UBR service enhanced with Early Packet Discard and also with a fair buffer allocation scheme. The results show that TOP performance over binary mode ABR is very sensitive to parameter value settings, and that a poor choice of parameters can result in ABR performance worse than that of the much less expensive UBR-EPD scheme.
- Research Organization:
- Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
- Sponsoring Organization:
- USDOE, Washington, DC (United States)
- DOE Contract Number:
- AC04-94AL85000
- OSTI ID:
- 385553
- Report Number(s):
- SAND-96-8547C; CONF-961125-1; ON: DE96009087; TRN: 96:005797
- Resource Relation:
- Conference: IEEE global communications conference, London (United Kingdom), 18 Nov 1996; Other Information: PBD: 27 Feb 1996
- Country of Publication:
- United States
- Language:
- English
Similar Records
Virtual queueing techniques for UBR+ service in ATM with fair access and minimum bandwidth guarantee
Effect of self-similar traffic on the performance and buffer requirements of ATM ABR edge devices