Google have launched a new service, called Google Public DNS. DNS, for those that don’t know, is the system that translates a domain name – such as www.esotechnica.co.uk – into the actual address that the computer connects to its services: such as retrieving a web page like this one, retrieving (or sending) your email, downloading a file, or just checking what time it is.

But before I talk about what Google Public DNS is and does, it’s worth reminding ourself what life was like before DNS…

Since 1969, when ARPANET (the predecessor of the Internet) first came into being, you needed to enter the addresses of the other computers you wanted to connect to into your local HOSTS.TXT file. This wasn’t particularly satisfactory, so after a while the Stanford Research Institute started to maintain a centralised list you could download (here’s an example from 1985), updated once or twice a week, with changes notified to them by email. Now imagine the size of file we’d need to maintain to list every computer on the Internet these days… completely unfeasible!

In November 1983, Paul Mockapetris published RFC 882, the first description of DNS. In essence, it described a distributed system, with root servers referring people to other domain name servers in turn to resolve queries. With the launch of DNS, it was possible for people to maintain their own IP address allocations, and for them to name and delete their own computers without having to refer that request to a centralised authority, and wait for everybody to download the latest file before the change could be effected. The switch to DNS was made universal less than a year later, in September 1984.

Twenty-five years later, Google’s analysis suggests that the DNS system, while still the backbone of the Internet, is beginning to prove a bottleneck. As might be expected – when DNS was invented, a 2 megabit connection to the Internet may have been prohibitively expensive, but these days it’s considered a relatively affordable consumer product.

Google’s Public DNS is essentially a research project. It’s a pair of caching DNS servers, that are open to the public, that aren’t authoritative on any domains (in other words, they only act as intermediaries), but are intended to be faster than using the ordinary DNS network. It’s not just providing lost of clustered, load-balanced servers that’s behind their approach, they’re also experimenting with new ways of working with DNS.

So…worth a look?

I’d say yet -  It’s intended to be faster than using the DNS network, and should reduce the load on other people’s DNS servers, rendering them faster too. So even if you don’t use Google Public DNS, the more people that do, the better for everyone. And as it’s a research project, it’s likely we’ll see some innovations from the project appear in RFCs, in BIND and other DNS servers.

But can it be used to track you, even if you’re not logged into Google? Kind of.  In a way.  It does give Google access to your DNS queries, but what would they do with them? The answer is – not much that can actually identify you personally. They could gather statistics on what domains are more popular, which could be used to improve Google Search, but since there is no user-based mechanism for Google to latch onto, it can’t easily be correlated with anyone’s search history (especially if you’re behind a firewall or using another public IP address.)

In short – it’s pretty cool, and worth supporting by using.  But it isn’t necessarily much different from your ISP running a caching DNS server, and there’s a small amount of usable information Google will get from you by your use of it.  Like everything else on the Internet, it’s your call, but personally?  Well, count me in…

Tags: ARPANet, DNS, Google, Google Public DNS, Paul Mockapetris, SRI-NIC

This entry was posted on Thursday, December 3rd, 2009 at 19:00 and is filed under News, WWW. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.