DNS Speed Comparison Report | Hacker News Reader

DNS Speed Comparison Report

solvedns.com

25 pointslox13 comments

we measure from Los Angeles, Dallas and New York

This is completely meaningless and borderline malicious (are they affiliated with one of the services that comes out on top?).

Real performance monitoring companies such as Cedexis, Gomez or Keynote measure from hundreds of locations, with a special focus on the 'Last Mile'.

The two metrics people tend to care most about (last mile and mobile performance in specific areas) are entirely unrelated to whatever ping-time a handful of vservers report.

It's a free chart with website/DNS related tools that are supported by ads.

If you have a similar service that offers free stats then please link to it.

If you don't do business in the US then this chart is probably of little use to you.

If you don't do business in the US then this chart is probably of little use to you.

Cedexis provides real data for free after registration; http://www.cedexis.com/radar/

If you don't do business in the US then this chart is probably of little use to you.

The chart is of no use to anyone, neither inside the US nor outside the US. It is worse than useless because it is actively misleading.

Spoiler: "we measure from Los Angeles, Dallas and New York"

Not very useful for European resident.

Agreed, and if it was just a blog post with some skewed stats it'd be fine. But it's made worse by the fact that DNS Made Easy are making advertising claims on the back of this [0]. This seems silly to me, because if SolveDNS introduce a Singaporean server next month, then DNS Made Easy will drop down the rankings [1].

[0] http://www.dnsmadeeasy.com/press-release/dns-made-easy-conti...

[1] Based upon some tests I just ran from a couple of different Singaporean servers. DNS Made Easy's closest PoP to Singapore is Hong Kong, with approx 35ms RTT. Cloudflare have a PoP in Singapore itself (<1ms RTT). Singapore was just chosen as an example, I'm sure there are many others.

I'd be interested in seeing how Google Cloud DNS ranks and how it's performance changes over time.

https://developers.google.com/cloud-dns/

sue.ns.cloudflare.com and theo.ns.cloudflare.com are below 2ms

ns1.google.com (through ns4) are 32ms I'm not sure if this is using their Cloud DNS service.

Does this change significantly throughout the day? On the main chart page it says Cloudflare averages 32ms.

Find out. Put a Google Cloud DNS hosted domain in the box in the upper-right of the page.

Now at 11am EST: (my original comment was at about 7am EST)

sue.ns.cloudflare.com = 48ms theo.ns.cloudflare.com = 240ms

ns1.google.com through ns4.google.com are all still at nearly exactly 32ms.

Interesting results...

I put in a vote for NSOne.net - which is not in the chart (They take issue with the testing from only Los Angeles, Dallas, and New York)

However, per: http://www.solvedns.com/nsone.net

After 4 lookups, the average response time of your name server dns1.p01.nsone.net is 11.56 milliseconds. After 3 lookups, the average response time of your name server dns2.p01.nsone.net is 11.91 milliseconds. After 3 lookups, the average response time of your name server dns3.p01.nsone.net is 11.60 milliseconds. After 2 lookups, the average response time of your name server dns4.p01.nsone.net is 11.61 milliseconds.

I'm not sure whether they do those lookups across locations or not, but that would put them at 2nd fastest in the list.

I think it matters what time you do the lookup. For Cloudflare it says the current response time is less than 2ms.

A couple of hours later the response time is anywhere from 40ms to 250ms. I guess that's what the monthly average is so important.

"The speed is the DNS lookup speed only, and does not include the name resolution speed. The time it takes to resolve a name server's name to its IP address is NOT included in the speed calculation."

Why NOT?

On any regular lookup, that data will be included in the glue records from the parent zone.

Are the averages means or medians? I'm guessing means and I am guessing this has skewed the output?

Reply on news.ycombinator.com