Optical interconnect can achieve a substantial increase in the number of nodes and switching capability for data centers, by virtue of their low power consumption and high bandwidth. In this paper, we propose a single-wavelength switch architecture based on two-stage AWGR for data centers. Then the single-wavelength design is extended to support multi-wavelength, which has higher scalability but lower hardware complexity and power consumption. We prove that the single wavelength architecture is internal strictly non-blocking. When the incoming traffic is externally blocked, a buffer allocation algorithm is designed to allocate feedforward and feedback FDL for the blocked packets. Based on the simulation results, the proposed switch can provide higher throughput, lower average latency, and a 0 out-of-order ratio compared to existing solutions.

With the rapid development of internet applications such as cloud computing and streaming media, people put higher demands on data center (DC) switching capabilities. Technology limitations of the backplane cause bottlenecks in its performance, which limits the growth rate of the communication ability of DC. The technology of waveguides based on polymer materials and waveguide connectors on the backplane has become an essential issue in the last few years. Architectures of optical interconnection remain to be explored. We propose a strictly non-blocking Clos-based optical architecture, which uses densely integrated optical waveguides to connect optical chips and achieve on-board networking. In addition, we design a Ring-Clos architecture to reduce the packet loss rate. The simulation results indicate that our design has good performance on throughput, packet loss rate, and latency.

The switching capability required by data centers is rapidly developing, however, the technical limitations of the optical devices have created bottlenecks in their performance. Additionally, architectures that support thousands of servers remain to be explored. With a distributed control, we present a high-radix architecture that exploits the space domain and the wavelength domain to provide high throughput, achieving fast configuration and low latency. Considering the network size, we also propose two methods for implementation. We firstly form a strictly non-blocking network under the constraint of one connection per port. To scale up the switch, wavelength division multiplexing (WDM) is utilized to interconnect more servers and achieve higher bandwidth. The simulation results indicate that the proposed architecture can scale over thirty thousand servers with throughput reaching 87% while providing low latency.