Modeling and Evaluation of Address Resolution Scalability in Data Center Interconnect Solutions

More and more services are provided by large data centers with a potentially very large number of hosts. The internal network in a data center is typically based on Ethernet. Different technologies exist in order to transparently interconnect distributed data centers. They typically tunnel Ethernet traffic over the wide area network. Two prominent examples for data center interconnection are Virtual Private LAN Services (VPLS) and the recently proposed Overlay Transport Virtualization (OTV). These two technologies use different mechanisms to disseminate Address Resolution Protocol (ARP) traffic, which can limit the scalability of large data center networks. This paper studies the control traffic caused by address resolution for these two data center interconnect technologies. We develop an analytical model for the ARP traffic between data center locations that takes into account the number of hosts and connected sites. With this model, the ARP traffic can be quantified both for VPLS and OTV. In addition, we study how an ARP proxy can improve the overall scalability, and we show that a proxy is beneficial for both technologies.