Comparative Study on the Development of Ethernet Flow Control

Ethernet has been the world's most comprehensive networking for more than 30 years. This popularity is due to its simplicity, maturity, low costs and a wide existing base. Fast and Gigabit Ethernet have brought more bandwidth to the technology and by transmitting from half-duplex to full-duplex link, Ethernet’s throughput has increased. With these properties Ethernet is expected to speed up the deployment and use of next generation networks and services. The current activity in building Gigabit speed networks, growing network bandwidth and intensive network applications led to flow control and congestion control became a high priority in network design and research. Recently, the IEEE congestion management study group has been set up to deal with the congestion and flow control problems in Ethernet networks. They had standardized a method (IEEE802.3x standard) to provide a pause-based flow control, which is applicable to all version of full-duplex Ethernet. The aim of this standard is to improve throughput, latency and packet loss. There are some problems for the IEEE802.3x standard; for instance, pause-based flow control can potentially cause heal-of-line blocking and more congestion, and in some cases quality of service (QoS) can not operate properly due to equal punishment for different traffic priority. Recent studies carried out some of these problems and have attempted to solve them in order to improve the IEEE802.3x standard. This paper analyzed and reviewed different proposed methods in terms of the strengths and weaknesses of their contributions. In addition, this paper provided future trends on the development of IEEE802.3x Ethernet flow control.