Switching in core optical networks is currently being performed using high-speed electronic or all-optical circuit switches. Switching with high-speed electronics requires optical-to-electronic (O/E) conversion of the data stream, making the switch a potential bottleneck of the network: any effort (including parallelization) for electronics to approach the optical speeds seems to be already reaching its practical limits. Furthermore, the store-and-forward approach of packet-switching does not seem suitable for all-optical implementation due to the lack of practical optical Random-Access-Memories to buffer and resolve contentions. Circuit switching on the other hand, involves a pre-transmission delay for call setup and requires the aggregation of microflows into circuits, sacrificing the granularity and the control over individual flows, and is inefficient for bursty traffic. Optical burst switching (OBS) has been proposed by Qiao, C., ?[1] to combine the advantages of both packet and circuit switching and is considered a promising technology for the next generation optical internet.
Performance Evaluation of High Speed TCP Variants over Optical Burst Switching Networks
