HIGH PERFORMANCE INTERCONNECTION NETWORK DESIGN USING CLOS ARCHITECTURE

Constant demand for more computing power has been a direct result of considerable growth in real-time applications, network virtualization and cloud computing. These bandwidth-starved applications and services of modern parallel computing are continuously driving the need for more powerful switching grids to manage the bulky packetized data traffic. These forward-looking computational powers are made possible by the interconnection of clusters of computers exchanging data that must provide fast and flexible communications at a reasonable cost. To achieve scalability and optimal performance requirements of such large switching networks, we employ a modular architecture that combines multiple stages of small-sized switches through crossbar connections. Clos architecture is implemented to for a highly scalable multistage interconnection network. Specifically, our focus is on the topology, routing, and flow control mechanism and how to guarantee network throughput for available input-output connections request within a node of a Switching fabric to set up connection paths without blocking. The principles of the Clos network are applicable to meet the requirements of today's networks; to create a highly performing interconnection IP fabric. Optimized Routing is the network conditions for nonblocking.