The top level server is either paid for by the domain owner (if they are managing the TLD name space internally), or by the company contracted to manage the TLD name space. This accounts for the top level servers in our diagram. What about the thirteen root servers? These are owned and operated by companies like Verisign (who manages two of them), and ISC, using funds collected from these same contracts. Other servers are managed by universities and research organizations, primarily through public and private grants and other sources of funding.
Let’s look at the operation of .com, managed by Verisign, specifically for a moment to get a better feel for how all of this works. Resellers purchase .com domain names from the TLD manager (TLDM in the diagram) for a fixed rate. No matter what a domain is actually worth, the reseller pays the same amount for it. Verisign has an entire front end system set designed to treat all the resellers of .com domain names, such as Network Solutions and GoDaddy, identically. For .com, the reseller makes money by capturing the differential between the fixed cost of purchasing the domain and what they actually charge for the domain. Names of little value are used as a “loss leader,” to convince people to use the resellers hosting plans or other services, and domains that are worth a lot (such as amazon.com) are sold for their intrinsic traffic gathering potential.
To understand the value of a domain name better, consider this scenario: You buy a domain name and build a successful social site with it, garnering a million hits a day. For some reason, you decide you’re not making money at it, so you dump the project and do something else. There are still a million users out there hitting that domain name every day — at least for a time, until they get used to seeing nothing there but a “site closed” message. Between the time you shut down and those users stop hitting your old web page, there’s an opportunity there for someone to take advantage of all those “free eyeballs” to build a new business. The reseller tries to take some of the value of those “free eyeballs” in reselling the domain to the next owner. It’s also common for some domain names to be “likely guesses” for users looking for a product. For instance, if you were looking for shoes, you might try “shoes.com” just on a lark. The reseller also tries to capture some of this value. In turn, these resellers actually own and operated the authoritative DNS servers.
Now we’ve figured out who pays for all but one of the servers in our diagram — who owns the recursive server we sent our first query to? There are a number of answers to this question.
Access providers own one set of theses servers, providing them as a part of their Internet access services. In this case, a local service provider owns and operates these servers. Individual companies also own and operate these types of servers as part of their normal internal IT operations. In either case, you’re paying for these servers as a customer of the internal IT department, a customer of the company with the internal IT department, or a customer of the local service provider.
Large providers, such as Google, also operate recursive servers. They pay for these servers through two means. The first is by improving the utilization of their networks by redirecting users the best place to enter their network, generally through load management. The second is by using the DNS queries as part of their overall data analytics program, allowing them to better target advertisements and provide a more complete picture of their customers to sell to others.
Companies like OpenDNS also operate recursive servers. They pay for these servers by building servers with more features (such as filtering content, or proxying requests and redirecting them based on device type to provide better service), and then charging to subscribe to services allowing access to these better servers.
There are some other uses for DNS that straddle the business and technical sides — we’ll cover these in the next post on HTIRW.