The goal of knowledge-center networking is to interconnect a giant number of server machines with low equipment price whereas providing high network capacity and high bisection width. It is well understood that this apply where servers are connected by a tree hierarchy of network switches cannot meet these requirements. During this paper, we have a tendency to explore a new server-interconnection structure. We tend to observe that the commodity server machines used in these days's knowledge centers usually come with 2 designed-in Ethernet ports, one for network connection and the opposite left for backup purposes. We believe that if each ports are actively employed in network connections, we have a tendency to will build a scalable, value-effective interconnection structure while not either the expensive higher-level large switches or any additional hardware on servers. We tend to style such a networking structure referred to as FiConn. Although the server node degree is only 2 in this structure, we tend to have proven that FiConn is extremely scalable to encompass lots of thousands of servers with low diameter and high bisection width. We have developed a low-overhead traffic-aware routing mechanism to boost effective link utilization primarily based on dynamic traffic state. We tend to have additionally proposed the way to incrementally deploy FiConn.
Did you like this research project?
To get this research project Guidelines, Training and Code... Click Here