Performance Evaluation Of TCP New Vegas And TCP Newreno On Burstification In An OBS Network

摘要

In a TCP over OBS network the TCP data is aggregated at the ingress node and the burst is formed based on time dependent mechanism or quantity dependent mechanism. In time based mechanism, the size of the burst may not be optimal and a small size burst may be formed which may cause wavelength contention to other burst. In quantity based mechanism the bursts are formed with a precise amount of data irrespective of the time needed to form the burst. In case of slow TCP sender and when the data used is non-real-time if we use quantity based mechanism, the TCP source may not generate data since the time of burstification is long and the data generated by TCP is waiting for burstification. This delay in aggregation of data is due to delay in generation of data by the TCP source. As a consequence, burstification cannot be done at a time and the OBS edge node is forced to wait for completion of burstification process. The delay caused due to burstification may influence the performance of TCP variants as it affects the calculation of RTT and therefore size of the congestion window. In this environment we have experimented with two important flavors of TCP, TCP Newreno and TCP New Vegas using NS-2. The results of the simulation prove that TCP New Vegas performs better than TCP Newreno when there is minimum delay due to burstification.