12.2 Internet 101: Understanding How the Internet Works

Host and Domain Names: A Bit More Complex Than That

While it’s useful to think of a host as a single computer, popular Web sites often have several computers that work together to share the load for incoming requests. Assigning several computers to a host name offers load balancing and fault tolerance, helping ensure that all visits to a popular site like http://www.google.com won’t overload a single computer, or that Google doesn’t go down if one computer fails.

It’s also possible for a single computer to have several host names. This might be the case if a firm were hosting several Web sites on a single piece of computing hardware.

Some domains are also further broken down into subdomains—many times to represent smaller networks or subgroups within a larger organization. For example, the address http://www.rhsmith.umd.edu is a University of Maryland address with a host “www” located in the subdomain “rhsmith” for the Robert H. Smith School of Business. International URLs might also include a second-level domain classification scheme. British URLs use this scheme, for example, with the BBC carrying the commercial (.co) designation—http://www.bbc.co.uk—and the University of Oxford carrying the academic (.ac) designation—http://www.ox.ac.uk. You can actually go 127 levels deep in assigning subdomains, but that wouldn’t make it easy on those who have to type in a URL that long.

I Want My Own Domain

You can stake your domain name claim in cyberspace by going through a firm called a domain name registrar. You don’t really buy a domain name; you simply pay a registrar for the right to use that name, with the right renewable over time. While some registrars simply register domain names, others act as Web hosting services that are able to run your Web site on their Internet-connected servers for a fee.

Registrars throughout the world are accredited by ICANN (Internet Corporation for Assigning Names and Numbers), a nonprofit governance and standards-setting body. Each registrar may be granted the ability to register domain names in one or more of the Net’s generic top-level domains (gTLDs), such as “.com,” “.net,” or “.org.” There are dozens of registrars that can register “.com” domain names, the most popular gTLD.

Some generic top-level domain names, like “.com,” have no restrictions on use, while others limit registration. For example, “.edu” is restricted to U.S.-accredited, postsecondary institutions. ICANN has also announced plans to allow organizations to sponsor their own top-level domains (e.g., “.berlin,” or “.coke”).

There are also separate agencies that handle over 250 different two-character country code top-level domains, or ccTLDs (e.g., “.uk” for the United Kingdom and “.jp” for Japan). Servers or organizations generally don’t need to be housed within a country to use a country code as part of their domain names, leading to a number of creatively named Web sites. The URL-shortening site “bit.ly” uses Libya’s “.ly” top-level domain; many physicians are partial to Moldova’s code (“.md”); and the tiny Pacific island nation of Tuvulu might not have a single broadcast television station, but that doesn’t stop it from licensing its country code to firms that want a “.tv” domain name (Maney, 2004). Recent standards also allow domain names in languages that use non-Latin alphabets such as Arabic and Russian.

Domain name registration is handled on a first-come, first-served basis and all registrars share registration data to ensure that no two firms gain rights to the same name. Start-ups often sport wacky names, partly because so many domains with common words and phrases are already registered to others. While some domain names are held by legitimate businesses, others are registered by investors hoping to resell a name’s rights.

Trade in domain names can be lucrative. For example, the “Insure.com” domain was sold to QuinStreet for $16 million in fall 2009 (Bosker, 2010). But knowingly registering a domain name to profit from someone else’s firm name or trademark is known as cybersquatting and that’s illegal. The United States has passed the Anticybersquatting Consumer Protection Act (ACPA), and ICANN has the Domain Name Dispute Resolution Policy that can reach across boarders. Try to extort money by holding a domain name that’s identical to (or in some cases, even similar to) a well-known trademark holder and you could be stripped of your domain name and even fined.

Courts and dispute resolution authorities will sometimes allow a domain that uses the trademark of another organization if it is perceived to have legitimate, nonexploitive reasons for doing so. For example, the now defunct site Verizonreallysucks.com was registered as a protest against the networking giant and was considered fair use since owners didn’t try to extort money from the telecom giant (Streitfeld, 2000). However, the courts allowed the owner of the PETA trademark (the organization People for the Ethical Treatment of Animals) to claim the domain name peta.org from original registrant, who had been using that domain to host a site called “People Eating Tasty Animals” (McCullagh, 2001).

Trying to predict how authorities will rule can be difficult. The musician Sting’s name was thought to be too generic to deserve the rights to Sting.com, but Madonna was able to take back her domain name (for the record, Sting now owns Sting.com) (Knorad & Hansen, 2000). Apple executive Jonathan Ive was denied the right to reclaim domain names incorporating his own name, but that had been registered by another party and without his consent. The publicity-shy design guru wasn’t considered enough of a public figure to warrant protection (Morson, 2009). And sometimes disputing parties can come to an agreement outside of court or ICANN’s dispute resolution mechanisms. When Canadian teenager Michael Rowe registered a site for his part-time Web design business, a firm south of the border took notice of his domain name—Mikerowesoft.com. The two parties eventually settled in a deal that swapped the domain for an Xbox and a trip to the Microsoft Research Tech Fest (Kotadia, 2004).

The Internet Is Almost Full

If you do the math, four combinations of 0 to 255 gives you a little over four billion possible IP addresses. Four billion sounds like a lot, but the number of devices connecting to the Internet is exploding! Internet access is now baked into smartphones, tablets, televisions, DVD players, video game consoles, utility meters, thermostats, appliances, picture frames, and more. Another problem is a big chunk of existing addresses weren’t allocated efficiently, and these can’t be easily reclaimed from the corporations, universities, and other organizations that initially received them. All of this means that we’re running out of IP addresses. Experts differ on when ICANN will have no more numbers to dole out, but most believe that time will come by 2012, if not sooner (Ward, 2010).

There are some schemes to help delay the impact of this IP address drought. For example, a technique known as NAT (network address translation) uses a gateway that allows multiple devices to share a single IP address. But NAT slows down Internet access and is complex, cumbersome, and expensive to administer (Shankland, 2009).

The only long-term solution is to shift to a new IP scheme. Fortunately, one was developed more than a decade ago. IPv6 increases the possible address space from the 2³² (4,294,967,296) addresses used in the current system (called IPv4) to a new theoretical limit of 2¹²⁸ addresses, which is a really big number—bigger than 34 with 37 zeros after it.

But not all the news is good. Unfortunately, IPv6 isn’t backward compatible with IPv4, and the transition to the new standard has been painfully slow. This gives us the equivalent of many islands of IPv6 in a sea of IPv4, with translation between the two schemes happening when these networks come together. While most modern hardware and operating systems providers now support IPv6, converting a network to IPv6 currently involves a lot of cost with little short-term benefit (Shankland, 2009). Upgrading may take years and is likely to result in rollout problems. David Conrad, a general manager at Internet Assigned Numbers Authority (IANA), the agency that grants permission to use IP addresses, has said, “I suspect we are actually beyond a reasonable time frame where there won’t be some disruption. It’s just a question of how much” (Arnoldy, 2007).

Some organizations have stepped up to try to hasten transition. Google has made most of its services IPv6 accessible, the U.S. government has mandated IPv6 support for most agencies, China has spurred conversion within its borders, and Comcast and Verizon have major IPv6 rollouts under way. While the transition will be slow, when wide scale deployment does arrive, IPv6 will offer other benefits, including potentially improving the speed, reliability, and security of the Internet.

But What If the DNS Gets Hacked?

A hacked DNS would be a disaster! Think about it. If bad guys could change which Web sites load when you type in a host and domain name, they could redirect you to impostor Web sites that look like a bank or e-commerce retailer but are really set up to harvest passwords and credit card data.

This exact scenario played out when the DNS of NET Virtua, a Brazilian Internet service provider, was hacked via a technique called DNS cache poisoning. Cache poisoning exploits a hole in DNS software, redirecting users to sites they didn’t request. The Brazilian DNS hack redirected NET Virtua users wishing to visit the Brazilian bank Bradesco to fraudulent Web sites that attempted to steal passwords and install malware. The hack impacted about 1 percent of the bank’s customers before the attack was discovered (Godin, 2009).

The exploit showed the importance of paying attention to security updates. A few months earlier, a group that Wired magazine referred to as “A Secret Geek A-Team” (Davis, 2008) had developed a software update that would have prevented the DNS poisoning exploit used against NET Virtua, but administrators at the Brazilian Internet service provider failed to update their software so the hackers got in. An additional upgrade to a DNS system, known as DNSSEC (domain name service security extensions), promises to further limit the likelihood of cache poisoning, but it may take years for the new standards to be rolled out everywhere (Hutchinson, 2010).