designing the DNS from the scratch

[Sten Carlsen]

On 09/07/2017 20:51, Reindl Harald wrote:
>
> Am 09.07.2017 um 20:41 schrieb Warren Kumari:
>> On Sun, Jul 9, 2017 at 1:59 PM John W. Blue <john.blue at rrcic.com
>> <mailto:john.blue at rrcic.com>> wrote:
>>
>>     Abdulhadi,____
>>
>>     __ __
>>
>>     Honestly, I think that a design spec of getting DNS responses in 3ms
>>     across the board is unrealistic.  My initial MX query for litc.ly
>>     <http://litc.ly> took 367ms:____
>>
>>     __
>>
>>
>> Like many poorly written / articulated SLAs, the devil is in the
>> details.
>>
>> I could happily read this as the server / service must respond within
>> 3ms. The OP mentioned VIP, so this could be for auth DNS, in which
>> case responding to a query within 3ms is trivial...
>
> no it is not - at least not if there is an internet connection between
> customer and dns server since you hardly get even a 3 ms ping time
>
> on the server itself yes
>
> frankly even in a local network you end with ;; Query time: 1 msec for
> a "dig NS" and that a nameserver can respond on localhost below that
> is completly worthless
>
Thinking about this, the only solution that seems remotely possible is
to let the customer have the DNS server on his premises. That way all
queries that come from cache could be fast enough. Those it needs to
resolve from the Internet, will take longer time.

The next question is who shall maintain that server on his premises? All
sorts of questions come to mind.

-- 
Best regards

Sten Carlsen

No improvements come from shouting:

       "MALE BOVINE MANURE!!!" 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.isc.org/pipermail/bind-users/attachments/20170709/e3e3187f/attachment.html>