[ntp:questions] More Granularity in the US in the NTP Pool

Danny Mayer mayer at ntp.isc.org
Sat Sep 15 17:37:33 UTC 2007


Henk Penning wrote:
> In <fcbsnu$e07$2 at usenet01.boi.hp.com> Rick Jones <rick.jones2 at hp.com> writes:
> 
>> I was wondering - is everyone confident enough that all these clients
>> choosing the "best" servers won't eventually settle on the same small
>> handfull of them?
> 
>   ... that could happen, but ... The pool-servers should be able to
>   cap the amount of traffic ; KODs for excessive clients and new clients
>   when the amount of traffic is "too high".
> 
>   This requires that 'demand' (by all clients) doesn't exceed 'supply'
>   (by all servers). If that happens, the pool project is doomed anyway.
> 
>   The goal of the poolclient project should be (in my view) a more
>   efficient match between supply and demand. Servers are more efficient
>   if they serve 'close by' clients. The poolclient mechanism allows
>   clients to find 'close by' servers.

This is more difficult than you make it sound even if you can come up
with a way of implementing this. The real problem is the large number of
badly written sntp clients who take no notice of RFC's, recommended
guidelines on best practice and the sheer number of these clients.

Danny


>> rick jones
> 
>   Henk Penning




More information about the questions mailing list