MCube: A high performance and fault-tolerant network architecture for data centers

A fundamental challenge in data center networking is how to efficiently interconnect a huge amount of servers. The network architecture typically consists of a tree of routing and switching elements with progressively more specialized and expensive equipment moving up the network hierarchy. Unfortunately, even when deploying the highest-end IP switches/routers, resulting topologies remain complicate application design and limit overall system performance. This paper presents MCube, a server-centric network architecture specifically designed for data centers. MCube leverages commodity mini-switches to provide high aggregate bandwidth and can deliver more performance at less cost than available from today's higher-end solutions. MCube is fully backward compatible with Ethernet, IP, and TCP and provides higher network capacity than the traditional tree-based structure for various types of services. Furthermore, MCube is also fault tolerant since it does not have single point of failure and its source routing protocol performs near shortest-path routing even in the presence of severe link or node failures. Experiences show that MCube is fault tolerant and load balancing and it significantly accelerates the bandwidth-intensive applications.