A Benhmarking Comparison of TCP Reno and TCP Vegas


Prajakta Pradeep Chakurkar

Oral Defence Date: 

Thursday, December 11, 2008 - 11:30


SCI 241


11:30 AM


Professors Marguerite Murphy, Dragutin Petkovic


The protocol suite used by Internet technology is called the TCP/IP protocol suite. TCP operates at the transport layer in the TCP/IP protocol suite. It ensures reliable and in order delivery of data. Therefore, it is frequently used by applications such as email, FTP, http (i.e. the WWW), all of which require guaranteed delivery of data. The Internet Protocol, IP, operates at the network layer and provides a “best-effort” service for delivery of data. But this may cause loss of important information in the network and/or out of order delivery of data at the receiver. TCP overcomes these disadvantages of IP by providing automatic reordering and retransmission of missing data. Due to increasing traffic over the Internet, network links may become bottlenecks. Many different variants of TCP have been introduced to improve TCP’s performance over high speed networks. These TCP variants help to avoid and control congestion in the network. TCP Reno was introduced in 1990 and is currently one of the most widely used variants. It provides a ‘fast retransmit fast recovery’ mechanism for congestion control. TCP Vegas was proposed in 1994. This protocol reduces packet drop in the network and is expected to provide higher throughput as compared to TCP Reno. The project presented in this report evaluates and compares the performance of TCP Reno and TCP Vegas. Our study includes designing and implementing a variety of benchmarking and network emulation experiments for these two TCP variants. The results of these experiments substantiate that, within the context of a single network link, TCP Vegas is a more effective congestion control algorithm, as compared to TCP Reno.

Prajakta Pradeep Chakurkar

TCP Vegas, TCP Reno, Gentoo Linux kernel 2.6.25-r7, Httperf, Autobench, ttcp, ECN, Netem.