We’ve written about DNS before - how important it is, how remarkably well it works, how often it’s attacked, and how taken for granted it is by the general public.

As such, last week’s announcement that Google is getting into the DNS business is worth paying some attention to (historically, DNS has been provided for at the Internet Service Provider level, but there’s no intrinsically technical requirement to do so).

Very briefly, the Domain Name System is a globally distributed database that maps names such as digitalmissive.com to IP addresses such 72.167. 131.220.  It’s important to realize that this is far more than merely a convenience to save web users (i.e. us humans) from having to memorize and type an IP address into a browser: a single website can make dozens of hidden ancillary calls involving other domain names (ad servers, traffic analytics, certificate authorities, etc.) during the process of loading, and (more importantly) the internet is about a lot more than web sites. 

It’s DNS that provides the vital layer of abstraction allowing for effective management of the myriad such connections and interdependencies upon which the internet as a whole depends.

Most of the media coverage of Google’s (still experimental) Public DNS program has had to do with concern over the company having control over yet another aspect of our online lives: the thinking is that Google already knows about your searches - use their DNS, and they’ll know about literally every site you visit.  For its part, Google has pledged to discard IP address information within 48 hours and to never correlate ISP and/or location data against any other Google account information - clearly, the company understands the ever-increasing importance of maintaining user trust as they look to expand beyond their current search-driven advertising business model into cloud computing.  While Google (despite some disconcerting recent remarks by CEO Eric Schmidt) has a pretty good record of responsibly handling such privacy matters, some feel that’s still an awful lot of personal information to be centralized in just one place.

However, I believe that Google Public DNS is less about the company getting (even more) intimately acquainted with our web browsing habits than it is about insuring a high level of performance and reliability for its upcoming Chrome cloud-based operating system.   Due sometime next year, the company is ‘throwing the long ball’ on this one - and they’re understandably not interested in leaving something as important as DNS to (ISP-dependent) chance.

So while concerns over Google Public DNS and web browsing privacy might be a bit of a red herring (I tend to trust Google more than my ISP anyway), what most analysis has tended to gloss over - and what I feel is the real story - is the technical implementation: rather than use common standardized server software such as BIND, Google’s system is entirely closed, opaque, and proprietary.  That’s the real paradigm change here - and yet another step in a long-term trend we’ve been following recently: The Privatization of the Internet.

Look at it this way: if the Chrome browser’s evolution from an application to an entire operating system (not to brag, but we saw that coming over a year ago) represents a transferal of client functionality upstream to the cloud, then Google Public DNS represents a transferal of ISP functionality (and responsibility) upstream to the cloud.

The sticking point for some is that it’s not ‘the cloud’ we’re talking about here - it’s ‘the Google cloud’.   However, to the extent the Chrome OS  will be fully could-based, DNS becomes the ‘glue’ holding the whole operating system together, perhaps a compelling technical argument could be made that a proprietary implementation  integrated within a proprietary system is warranted - even at the possible expense of privacy.