Testing performance of local DNS lookups

Re: Testing performance of local DNS lookups

From:

Date:

Time:

Think of it this way:
Your DNS server possibly is only doing outside lookups for a small
group of users (maybe just you), so if you ask it for something, and
it has not look up that domains information before, or recently, it
will have to contact the root DNS servers, find out the DNS servers
handling that domain, and then request their dns servers to provide it
the needed information, then pass that back to you.

Comcasts DNS servers have a much larger userbase looking up domain
information, so probably already have the required information cached
locally for the TTL value set in the domain in question. So you
request it from their DNS servers and they send it right back to you,
no need to go out and get the information from two other sources
first.

Now when you are doing your dig testing are you using domain names you
have already recently visited? Try picking a domain name you know
probably is not in your DNS servers cache, but is probably popular
enough that Comcast would have had someone do a visit to.

You will probably notice a bigger difference in lookup times. If you
then turn around and do the same lookup again with dig they will
probably be much closer as your DNS server now as that information
locally for the TTL value for that domain.

Hope that makes sense, I have not even started my first cup of coffee
yet.

On Tue, 26 Oct 2004 22:55:39 -0700
Scott Haneda <lists@newgeo.com> wrote:
>And my saga continues...
>
>Forever I have had my DNS servers IP listed in my tcp/ip settings,
>this has
>served me well for ages now. After experiencing some DNS issues, I
>decided
>to enter in comcasts DNS servers in my tcp/ip settings. It is like a
>whole
>new web experience. Where I was used to a certain delay in resolving
>a
>domain name before Safari would display it, now it is immediate.
>
>How can I test where this bottleneck is happening?
>
>So far, I have done this:
>Dig somedomain.com @my.ip.address
>Dig somedomain.com @comcast.dns.ip.address
>
>Both results come back in about the same ms time range, give or take
>1-5
>ms's. Is this the proper way to test this, these results seem to
>lead me to
>think that I should have the same experience on both ip addresses,
>which is
>certainly not the case.
>
>Maybe I have been going about this all wrong, and should not have
>used my
>own DNS server for my lookups and should use that of the provider
>that is
>giving me my internet? I just find it hard to believe that with as
>many
>people as comcast has on the system, that little old me can not have
>a
>faster DNS server than them considering load.
>
>I really don't want to use another NS than mine anyway, since I want
>to see
>the edits I made to DNS right away, which helps me in development and
>the
>like.
>--
>-------------------------------------------------------------
>Scott Haneda Tel: 415.898.2602
><http://www.newgeo.com> Fax: 313.557.5052
><scott@newgeo.com> Novato, CA U.S.A.
>
>
>

Sales:
888/4-DPOINT

Support:
858/452-3696