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

Henk Penning henkp at cs.uu.nl
Mon Sep 10 16:50:43 UTC 2007


In <jhqcr4-499.ln1 at klein-habertwedt.de> Uwe Klein <uwe_klein_habertwedt at t-online.de> writes:

>evandro at mailinator.com wrote:
>> I would however add that having time-zone aliases would make it easy
>> for OS distributions to automatically configure the NTP servers.  For

>The next best thing imho would be to select by lowest hopcount and ignore
>any geographic details.
>
>In internet space earth is not flat. It has lots of folds, wormholes
>and anisotropic path.

So, ... wouldn't it be nice if ntpd would solve this problem!

Suppose a ntpd client can pick up 5 random pool servers, and
periodically (say once a day) replace the 'worst' server
by randomly picking a new one.

A little simulations shows that the client would quickly evolve
to a stable best-3-of-5 set of servers. Every client would slowly
drift to using a set of 'good' servers.

-- The clients are happy because they have 'good' servers
-- The servers are happy because the have good clients (low poll rate)
-- no hasle with complicated geo-aware DNS schemes

Of course the 'asynchronous dns lookup' problem has to be solved first,
but when that's done, the scheme above would work nicely, I think.

The protection for the server (too many clients, bad clients)
is in place (rate limits, KODs etc).

Henk Penning

-- 
----------------------------------------------------------------   _
Henk P. Penning, Computer Systems Group       R Uithof CGN-A232  _/ \_
Dept of Computer Science, Utrecht University  T +31 30 253 4106 / \_/ \
Padualaan 14, 3584CH Utrecht, the Netherlands F +31 30 251 3791 \_/ \_/




More information about the questions mailing list