[ntpwg] [dhcwg] NTP option: IP address and/or FQDN

[Danny Mayer]

Richard Gayraud (rgayraud) wrote:
> Hello Alain,
> 
> This is not really different, but the impact of it will be less
> important:
> 
>   - If an hardcoded NTP IP address is shipped within a SOHO thing,
>     (as it happened in the past), then this IP address is dead, 
>     for good. Not usable anymore. Also, I suspect routers on the
>     path to this subnet are impacted, unless global internet
>     routing tables are updated to drop this traffic upstream.
> 
>   - on the other side. If a DNS name is hard-coded, this IP 
>     address and subnet issue can be avoided by removing the
>     name from the DNS database. 
> 
>       . Clients are not supposed to re-query the DNS server
>         every 2 seconds after the server replies the name is
>         not resolvable.
> 
>       . even if they do, the load will be distributed over
>         multiple local DNS servers (instead of having a
>         single victim IP address).
> 
>       . we hope that a poorly coded SOHO equipment will use a 
>         pool FQDN rather than a single NTP server name.
> 

There is an even simpler solution to this. Have the DNS name remain but
have it point to 127.0.0.1 or ::1. That way only itself is impacted.
Normally one isn't supposed to do this but in this case it's good
solution and solves just about every problem and does not affect other
nodes in the network.

Danny

> I think other people gave other good reasons to use FQDN, but
> I do not remember all of them.
> 
> Does this help ?
> 
> Thanks,
> 
> Richard.
> 
>> -----Original Message-----
>> From: Alain Durand [mailto:alain_durand at cable.comcast.com] 
>> Sent: dimanche 9 décembre 2007 19:55
>> To: Richard Gayraud (rgayraud); Danny Mayer; MORAND Lionel RD-CORE-ISS
>> Cc: DHC WG; Ralph Droms (rdroms); NTP Working Group
>> Subject: Re: [dhcwg] NTP option: IP address and/or FQDN
>>
>>
>>
>>
>> On 12/9/07 10:16 AM, "Richard Gayraud (rgayraud)" <rgayraud at cisco.com>
>> wrote:
>>
>>>   => But more importantly, this does not give any guaranty 
>> that a vendor
>>>      will not ship a small home router with a DHCP server 
>> inside, with an
>>>      embeded NTP server address => worst case.
>> Excuse me if I'm just adding fuel to the fire, but I still 
>> fail to see the
>> difference between a vendor shipping a home router with a 
>> DHCP server that
>> has an NTP option embedding a hard-coded IP address and the 
>> same vendor
>> shipping a similar product with a hard-coded FQDN...
>>
>>    - Alain.
>>
>