Among Google’s recent announcements was their introduction of Google Public DNS. I’ve been using OpenDNS and have no complaints. Well, actually I recently found I had defaulted back to using my ISP’s DNS (Comcast), probably during a router firmware upgrade. When I switched to back OpenDNS I also didn’t notice a different over Comcast. I wouldn’t have noticed unless I was in the router config for another reason and happened to see it.
Comcast and OpenDNS both do typo hijacking and display a search page with ads rather than an error page. I went through the process of opting out of Comcast’s typo hijacking. OpenDNS also allows an opt-out for typo hijacking which I have set. Interestingly enough, the advertising company – Google, doesn’t hijack typos for ads and they display the error page for typos. But this lack of hijacking wasn’t a benefit for me since my opt-outs were already in place and were working fine.
To be honest I didn’t notice any performance difference when I was set to use any of them. When I first switched from Comcast to OpenDNS long ago I did notice imroved performance, but not this time. So I went looking for a way to benchmark performance and came across namebench. It’s simple to use and provides useful information.
Just download namebench and run the executable. You’ll be presented the following screen:
The “Benchmark Data Source” is a drop down that let’s you pick one of your browsers or the Alexa Top Global Domains as a data source. Picking your most used browser provides results that are specific to the way you browse. Some people have complained that this could send all your browsing history to one person (the Google developer). Since the source code is public it’s easy to confirm it doesn’t. But, if your still concerned, picking Alexa will use generic sites.
Click “Start Benchmark” to get things going. Once the benchmarking is done (took about 10 minutes for me) a page with the results will open in your browser. At the top will be the information yiu really want:
The above result is from a run after I’d already re-configured for it’s previous recommendations and OpenDNS is the second fastest DNS server according to the benchmark. The right box displays the recommended DNS servers that should be used. In my case the first one is the internal IP of my local router so should be ignored. (I didn’t include it in the screenshot but you’ll get detailed info on the servers tested. See the previously linked namebench page for samples.
The bottom line is Google Public DNS didn’t make the cut. So, while the accuracy of the benchmark may be questioned (as would any benchmark) it’s pretty clear there’s no Google favoritism. M5Net, UltraDNS and Comcast were my recommended DNS servers. Another note, because of caching the first time run of namebench will deliver the most accurate results.
So, I started off by looking at Google Public DNS but by the time I was done I was off of it. But looking into it I considered the following:
- Google doesn’t have to match the info to me to benefit. The additional information they collect about were people surf and how often is a treasure waiting to be mined. They don’t need to put ads on error pages to profit from DNS.
- Google does continuously hit on speeding up the web so it’s likely they’ll keep improving performance. They have studies showing that slow response on their search results generates lower revenue.
- They also promote security and Google certainly has the money and talent to keep DNS as secure as possible.
Like my recent foray into Google’s Picasa/Eye-Fi deal, Google Public DNS is yet another Google offering that sounded good but wasn’t quit right for me. Like Picasa, Google DNS will stay on my radar and I’ll check it out sometime down the road. Anyone else trying Google Public DNS?