How Fast Is Your ISP’s DNS Server; Compare with Speed of OpenDNS

In your example, the times are in HH:MM:SS.SS, so the differences are in seconds. That makes youre differences roughly 10 seconds and 4 seconds. This isn’t an accurate test though since the first prompt of each test was displayed before you even typed the commands, not as the command was executed.

To make this a little more accurate, you could just created a basic batch file with the two nslookup commands, then execute the batch (leaving echo on). It’s output for you test would look more like this:

8:51:01.02dnstest

8:51:03.52nslookup link 208.67.222.222
Server: resolver1.opendns.com
Address: 208.67.222.222:53

Non-authoritative answer:
Name: link
Address: 208.67.217.130

8:51:03.63nslookup link 125.22.47.125
Server: AES-Static-125.47.22.125.airtel.in
Address: 125.22.47.125:53

Non-authoritative answer:
Name: link
Addresses: 209.85.175.104, 209.85.175.99, 209.85.175.147
Aliases: link

8:51:07.34

In that example, 8:51:01.02 was the time that my initial prompt was displayed. 8:51:03.52 was when the query against OpenDNS started. It then finished and executed the command against your DNS server at 8:51:03.63. The second lookup finished and the batch file was complete at the time of the final prompt, 8:51:07.34. Computing that, we get:

OpenDNS: .11 seconds / 110ms
Airtel: 3.71 seconds / 3710ms

Actually, I also have Airtel as my local ISP, and I get different results. Also, in my experience Google is the worst test case for tests like this (as most places at least Google will always be cached).

I tested various sites with this method (msdn.com, yahoo.com, etc.); in some cases OpenDNS was better, and in others, Airtel.

I like to use OpenDNS also for the fact that I don’t have to worry about the DNS wherever I go. I don’t ever change my DNS.

how is the difference calculated here? is it not the difference between the time last time enter was pressed and when the command was executed?

Amit, either its a misunderstanding on my part of its an error in your statement.
Lemme explain -

1. When you type prompt $t and the the prompt which appears, it shows the time of the appearnace - now to that when you type the command and execute it, there can be a huge time difference based on when you executed and when the initial prompt had come.
Lemme just explain with what you did and where your assertion seems wrong -
1. 17:45:49.62 is the time when the prompt $t executed and appeared and 17:45:59.43 is when the nslookup got completed and time prompt appeared again.
First error - its 17 hrs, 45 minutes, 49.62 seconds. Now, when you executed prompt $t and lets say leave it for a minute or so and execute the nslookup - it’ll show a time difference of 1 minute. You have wrongly asserted it to be 10ms seeing digits 49 and 59 which are actually representing 10 seconds difference which could be time taken to type the entire nslookup command including execution.
Second case you found it to be 4 and you said 4ns - Its kinda misleading(milli to nano) when it is all dependednt on how soon you entered the two commands and pressed the “Return” key.

Am using the upper arrow key to get the previous command and pressing “return” - well, Airtel seems faster for sure as I see a 2 second vs <1second execution time difference. Actual is what the script may give :)

Buts its definitely not the ‘ms’ and ‘ns’ numbers you gave.

Thanks
I think it’ll be perfect if these commands are written in a script and executed - not sure if it’ll be right even then.

I hope you got my point.
Else, feel free to write back :)

Just use dig which is much better at showing timing information.

Does this really work? My DOS prompt shows time in hh:mm:ss.ss format. Assuming the same format applies in the example then it took 10 seconds for OpenDNS and 4 seconds for Airtel. I think this is because the time includes the time to type the command as well as execute the command.

Thanks for taking a look at OpenDNS.

I’d point out that the benefits of OpenDNS’s huge caches are more notable when checking any domain that is not the world’s most popular search engine. ;-) google.com is likely to be in any/all caches, no matter what DNS you’re using. Stray beyond the obvious, and the benefits of OpenDNS’s huge caches start to stack up.

Also, consider the reliability of a zero-downtime, globally-deployed, anycast network.

Just saying…

John Roberts
OpenDNS

Amit, I just posted a (what I believe) comprehensive test on OpenDNS versus Airtel.

link

If you want, you can point your readers to that.

Cheers.

Dear Mr.Amit,

The Time difference what we see is actually the time taken for us to Type

Only if you run a batch command you will know the actual time

Regards
Bruno

To do this in unix/linux/macosx, type “time” before the nslookup command. Such as…

time nslookup google.com 208.67.222.222

Where you will get results like..

———
$ time nslookup google.com 208.67.222.222
Server: 208.67.222.222
Address: 208.67.222.222#53

Non-authoritative answer:
Name: google.com
Address: 64.233.187.99
Name: google.com
Address: 209.85.171.99
Name: google.com
Address: 72.14.207.99

real 0m0.022s
user 0m0.003s
sys 0m0.006s
——-
Note that in this case, it is in full seconds, so you have to move the decibel - 0.022seconds = 22 milliseconds.

yeah man amit that testing proceedure is useless. even with a script or batchfile its so not accurate.
eventhough.. lookuptime is in .ms… it also depends just how fast your dosbox responds to the cmd. also u need to test an ip a few times to get an average.
only a realtime app will give u accurate results that u can use to compare speeds. also those results can change the next day. unless u have a problem theres no meaning to test compare time…