A high performance network architecture for large-scale cloud media data centers

摘要

Cloud media services will become indispensable to enrich our future daily life. However, bandwidth-intensive cloud media services involve huge amount of data traffic while the requirements for quality of services (QoS) are diverse among different clients. Conventional data center networks (DCN) ¹ meet difficulties in terms of scalability and flexibility to support the increasing bandwidth requirements for cloud media services. To solve this problem, a new type of architecture, MatrixDCN, is proposed in this paper. MatrixDCN is an approximate non-blocking network, in which switches and servers are arranged in rows and columns that compose a matrix structure. A MatrixDCN based network can accommodate up to hundreds of thousands of servers with good scalability. A high-efficient routing algorithm is well designed to fully leverage the regularity in the topology while computing a routing table in a simple way. Theoretical analysis and experimental data show that MatrixDCN have advantages on topology's scalability, network throughput, and its routing algorithm.