Simple, efficient location‐based routing for data center network using IP address hierarchy

Increasing demand and sophistication of applications deployed on data centers resulted in various designs for data center networks (DCNs). One of the major challenges in the design of DCNs is the design of routing protocol that scales to support millions of servers that a typical DCN hosts. Many alternative routing protocols are proposed to overcome the scalability problem of conventional routing protocols such as Open Shortest Path First and Routing Information Protocol. These alternative protocols that use topology characteristics of DCN are broadly classified as source routing and location‐based routing. In the process of fixing the scalability problem, these protocols introduced additional complexities such as large network control overhead and reprogramming of network elements. The extra control overhead in these protocols is the result of their effort to determine the relative location of the end hosts in a given topology. Further, existing location‐based routing is not entirely location based and covers only the latter half of a route. In our work, we present a new location‐based routing based on IP address hierarchy that (a) does not need any additional network control plane and management planes, (b) deployable on proven network technologies, and (c) covers entire path of the route. We establish the correlation between topology design and address assignments that helps determining the location of an end host directly from the address assigned to it. We demonstrate our proposed location‐based routing on an existing proven architecture for DCN, BCube‐IP and on our proposed architecture 4‐4, 1‐4. We give proper justification for proposing 4‐4, 1‐4, a better design for our proposed location‐based routing. Copyright © 2016 John Wiley & Sons, Ltd.