New on LowEndTalk? Please Register and read our Community Rules.
What's your DNS Hosting Speed?
I used to hate when IE would take a long time to look up the website url. So over the years I have tested various DNS setup. For me dnsmadeeasy.com is the winner. Here is a quick comparison of three DNS providers for three of my domains.
I used this site for the test. https://www.ultratools.com/tools/dnsHostingSpeed
- dnsmadeeasy.com
- Amazon Route 53
- DNS.HE.NET
Every little second count, right? 
What's your speed? 
dnsmadeeasy
!(http://gio.myterabyte.com/myimages/forum/dnsmadeeasy.jpg)Amazon Route 53
!(http://gio.myterabyte.com/myimages/forum/aws.jpg)DNS.HE.NET
!(http://gio.myterabyte.com/myimages/forum/dnshe-2.jpg)
Comments
The Ultratools DNS speed test is fairly useless (unless the majority of your site's users are located in Virginia) because the results it is giving are the ping times from the ultratools.com website (datacenter location: Virginia) to the servers of the DNS providers.
Additionally, it doesn't actually seem to request a record? I know that it says AAAA, CNAME, etc at the top, but I tried it against our own DNS cluster, then looked at the logs to see where the requests were coming from, and...there's nothing in the logs?
Can anyone else confirm this?
=======
Additionally, everywhere I tried was less than 100ms, which is pretty imperceptible. Regarding this:
This would probably be more of an issue with your ISP's DNS servers than site's DNS hosting.
I can confirm it didn't make any DNS queries to our nameservers. The only thing it seems to do is a ping test of the nameserver's IP.
It is a great marketing tool for UltraDNS though because their DNS service will always be #winning when you test with this tool.
Test akamai.com and you'll see some times in the 180-240ms range for their nameservers in Europe and Asia (the akamai nameservers in Ashburn Virginia have times in the 1-2ms range)
Rage4 DNS Case Study: tr-ance.com - http://gbshouse.com/promo/Rage4CaseStudy_tr-ance.pdf
http://www.solvedns.com/blog/major-dns-providers-speed-comparison/
1) HE.NET
2) CloudFlare
3) Dyn
@gbshouse
why don't you use your Rage4 DNS for your main domain rage4.com?
@gbshouse, do you recommend a web tool to actually test DNS server lookups?
Might try http://just-dnslookup.com/ ?
i use dnsmadeeasy as well for dns for my more important domains (on their business plan)
@kornnflake - multiple reasons, for example: we want to use two separate networks or we don't want to change current setup
@pubcrawler - for lookup diagnostics? this one
Anyway the DNS hosting speed is a little bit hard to measure. First of all I would assume that ping response time is not the best metric, beside that for records query you should use low TTL (equal to 1) so you will be sure that records is served from provider itself not from recursive/public DNS (like Google or OpenDNS)
@DomainBop For the dns.he.net name server, the website is in Los Angeles. For the Amazon Route 53 name server, the website is in Lansing, Michigan. For dnsmadeeasy.com, the server is in Phoenix, Arizona.
I have traveled to Europe and Asia and the dns speed result is the same for the dnsmadeeasy.com name servers.
Could you not just run:
Gives you the time in ms.
Why not just
If you want to test your dns server directly (bypassing any in-between caching servers).
I am wondering with @sleddog's example the value of DNS lookup timing solely on your own nameserver.
Better real world testing would involve popular DNS servers as test points (4.2.2.1, 8.8.8.8, 208.67.220.220)
dig @4.2.2.1 www.yourdomain.com
dig @8.8.8.8 www.yourdomain.com
dig @208.67.220.220 www.yourdomain.com
But @pubcrawler if you do this:
dig @8.8.8.8 www.yourdomain.com
then you're just testing Google's anycast network of caching name servers. @sleddog is correct if you run your own name servers and want to test response times for your own domains, although you'll have to run that test from various locations to get a better idea.
You could do a test through the public caching servers like Google's if you create a new DNS record that you know has never been cached so Google's server would have to query your own name servers first before returning a result.
Yes. That's what I meant. I should've clarified.
I don't see much point to testing public caching servers as there's nothing we can do about it if performance sucks