I Guess there's something to say for both sides: I agree with Tim that
it's probably to considered "progress" that they're now using the
pool, in stead of hitting on one unlucky stratum 1 somewhere in the
world... ;-)

And as long as their client behave's alright, nothing's really
bothering anyone as the pool is just being used as it was intended: to
spread the load. But then again: if they don't, it'll be hard to tell
where shit is coming from when it suddenly does hit's the fan.

And if someone out there came as far as learning that there's a pool
out there, why not go the extra mile, unless they're to cheap to
donate and don't want to deal with the guilty feeling  of asking for a
vendor-pool.

And Sam Mason explains the rest: probably, someone taking enough care
to find out about wich NTP to use and what the rules of engagement
are, probably took care of a decent NTP-implementation as well. As for
someone who doesn't care about the one thing, probably won't about the
other.

Roelant.

On Mon, Nov 23, 2009 at 6:16 PM, Scott Baker <[email protected]> wrote:
> On 11/23/2009 09:14 AM, Shoppa, Tim wrote:
>>
>> I think that because of the way the load will be distributed across time
>> and space, that the
>>
>> vast majority of vendors who use well-behaved NTP clients should be
>> allowed as part of
>>
>> the rules of engagement to use pool.ntp.org without even asking.
>>
>> It’s the boneheaded clients that request at the top of every hour, or
>> that start
>>
>> Launching requests at insane rates on certain return codes, that are the
>> problem children.
>>
>> These still exist but aren’t quite as bad as a few years ago IME.
>
> That's all well and good, but how do know that Vendor XYZ is going to
> "behave". I think if a vendor wants to deploy a large amount of NTP devices
> they should have their own XYZ.pool.ntp.org dns entry. That gives the
> vendor, and us more control of how and where those packets go.
>
> Also, it's trivial to setup. Not to speak for ASK, but setting up a DNS
> entry should take about 15 minutes.
>
> - Scott

_______________________________________________
timekeepers mailing list
[email protected]
https://fortytwo.ch/mailman/cgi-bin/listinfo/timekeepers

Reply via email to