On the problem of optimization of DNS root servers' placement

Tony Lee, Brad Huffaker, Marina Fomenkov, and kc claffy
Cooperative Association for Internet Data Analysis - CAIDA
San Diego Supercomputer Center,
University of California, San Diego

The Domain Name System (DNS) is a critical component of the modern Internet. It provides a critical link between human users and Internet routing infrastructure by mapping host names to IP addresses. The DNS is a hierarchy of distributed system of servers anchored at 13 DNS root servers.

In this paper we examine the macroscopic connectivity between the DNS root servers and the worldwide population of their clients. We study the impact of the geographical locations of root servers on the latency of server-client connections. We also propose a methodology to estimate the effects of root servers' relocation.

We found that all root servers can be clustered in four groups that closely correlate with their geographical positions. Servers in the same group are nearly indistinguishable for their clients in terms of latency and can replace one another in providing DNS services to the clients. M-root, the only root server in Asia, is in a group of its own and, therefore, is the most crucial for its clients in terms of the latency increase in case of its unavailability. Clients in Europe appear to be relatively underprovisioned and may merit an additional root server. Clients in North America appear overprovisioned. One of the US servers may be a suitable candidate for relocation to a different region of the world.