TCP-Vegas Slow Start Performance in Large Bandwidth Delay Network
نویسندگان
چکیده
With the rapid expansion of the Internet, it has become possible for end hosts that are separated long apart to be connected through high bandwidth links. This environment, called a Large Bandwidth Delay Network, poses a major challenge to the performance of the Internet. A long-delay connection usually suffers from being treated unfairly when competing with short-delay connections. A link, to avoid being under-utilized, has to be equipped with an a buffer as large as the bandwidth delay product of the longest connection. TCP-Vegas is known as a potential solution to these problems. According to a number of previous studies, it is said to fairly treat connections with different propagation delays and avoid under-utilization even with a buffer that is independent of the bandwidth delay product. In this paper we show, from simulation and analysis, that the current TCP-Vegas does NOT achieve high utilization in such a large bandwidth delay network, because of its slowstart phase. Moreover, to avoid loss, TCP-Vegas slow-start requires a buffer that is proportional to the square root of the bandwidth delay product. We propose a solution to these problems and analyze its performance. Keywords— TCP, congestion control, TCP-Vegas, slowstart, Large Bandwidth Delay, LBD, utilization, convergence delay, burstiness.
منابع مشابه
On the transient behavior of TCP Vegas
Research has shown that TCP Vegas performs better than TCP Reno with respect to overall network utilization, stability, fairness, throughput, packet loss, and burstiness. In this paper, we analyze and improve the transient behavior of TCP Vegas, an important issue in today’s large “bandwidth-delay product” networks. To quantify of our analysis, we introduce a new metric that captures the transi...
متن کاملTCP Start up Performance in Large Bandwidth Delay Networks
Next generation networks with large bandwidth and long delay pose a major challenge to TCP performance, especially during the startup period. In this paper we evaluate the performance of TCP Reno/Newreno, Vegas and Hoe’s modification in large bandwidth delay networks. We propose a modified Slow-start mechanism, called Adaptive Start (AStart), to improve the startup performance in such networks....
متن کاملA Logarithmic Slow-Start Algorithm of TCP Vegas in IP networks
In numerous TCP congestion control algorithms, TCP Vegas avoids congestion on its own initiative. However, its slow-start ending quickly and growing unsmoothly cause that the congestin window designing too small and bandwidth utilization declining too rapid, furthermore affect network efficiency and throughput dropping. . . etc. Aimed at the deficiencies, we propose a kind of Logarithmic Slow-S...
متن کاملImproving TCP Start-up Behavior in High-Speed Networks
TCP is a reliable data transfer protocol used widely over the Internet for numerous applications, from FTP to HTTP. The current implementation of TCP Reno/NewReno mainly includes two phases: Slow-start and Congestionavoidance. In Slow-start phase, the sender opens the congestion window (cwnd) exponentially, doubling cwnd every Round-trip Time (RTT) until it reaches the Slow-start Threshold (sst...
متن کاملVEGAS RB+: An Optimized Version of TCP VEGAS
TCP Vegas is a well-known delay-based congestion control mechanism. A strategy called Additive Increase/Multiplicative Decrease regulates the number of packets that are sent at one time. When congestion occurs the packets start to drop (multiplicative decrease). TCP's four intertwined congestion control algorithms are: Slow Start, Congestion Avoidance, Fast retransmit, and Fast recovery. In thi...
متن کامل