[131453] in North American Network Operators' Group
RE: NTP Server
daemon@ATHENA.MIT.EDU (Brandon Kim)
Sun Oct 24 20:16:04 2010
From: Brandon Kim <brandon.kim@brandontek.com>
To: <sean@donelan.com>, nanog group <nanog@nanog.org>
Date: Sun, 24 Oct 2010 20:15:56 -0400
In-Reply-To: <alpine.GSO.2.00.1010241348220.2347@clifden.donelan.com>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org
Hi Sean:
By local I meant in-house=2C on-site in our datacenter. As far as what appl=
ications could use our NTP service=2C I would
leave that up to each client and what they are running. For my own personal=
purposes=2C it would just be for log purposes.=20
(error logs=2C syslogs=2C etc etc)
I have heard that routers don't make good NTP servers since they weren't de=
signed to keep track of time. This=2C I have read
from a Cisco source. Can't remember where though. Or maybe they were just r=
eferring to older less powerful routers like 2500 series...
Brandon
> Date: Sun=2C 24 Oct 2010 14:42:24 -0400
> From: sean@donelan.com
> To: nanog@nanog.org
> Subject: Re: NTP Server
>=20
> On Sun=2C 24 Oct 2010=2C Brandon Kim wrote:
> > 1) How necessary do you believe in local NTP servers? Do you really=20
> > need the logs to be perfectly accurate?
> > 2) If you do have a local NTP server=2C is it only for local internal=20
> > use=2C or do you provide this NTP server to your clients as an added=20
> > service?
> > 3) If you do have a local NTP server=2C do you have a standby local NTP=
=20
> > server or do you use the internet as your standby server?
>=20
> First terminology. What do you mean by a local NTP server?
>=20
> Almost any Cisco/Juniper router=2C Unix server and some recent Windows=20
> servers have NTP server software and can synchronize clocks in your=20
> network. So you may already have a NTP server capable device. You just=
=20
> need to configure it=2C and give it a good source of time. It would be a=
=20
> Stratum 2 or greater NTP server because the good source of time is=20
> another NTP server. Left to itself=2C NTP is pretty good at keeping cloc=
ks=20
> in arbitrary networks synchronized with each other. But most people are=20
> also interested in synchronizing clocks with some official time source.
>=20
> The Network Time Protocol doesn't really have the notion of a "standby"=20
> server. It uses multiple time sources together=2C and works best with ab=
out=20
> four time sources. But for many end-systems=2C the Simple Network Time=20
> Protocol with a single time source may be sufficient.
>=20
> If you are in a regulated industry (stock broker=2C electric utility=2C 9=
-1-1=20
> answering point=2C etc) there are specific time and frequency standards y=
ou=20
> must follow.
>=20
> On the other hand=2C are you are asking about a local clock receiver (rad=
io=2C=20
> satellite=2C etc) for a stratum 1 NTP server? Clock receivers are gettin=
g=20
> cheaper=2C the problem is usually the antenna location.
>=20
> Or on the third hand=2C are you asking about local primary reference cloc=
k=20
> (caesium=2C rubium=2C etc) for a stratum 1 NTP server? These are still=20
> relatively expensive up to extremely expensive.
>=20
> Or on the fourth hand=2C are you a time scientist working to improve=20
> international time standards. If you are one of these folks=2C you alrea=
dy
> know.
>=20
>=20
> Most major ISPs use NTP across their router backbone=2C and incidently=20
> provide it to their customers. The local ISP router connected to your=20
> circuit probably has NTP enabled.
>=20
> Required accuracy is in the eye of the beholder. NASDAQ requires brokers=
=20
> to have their clocks synchronized within 3 seconds of UTC(NIST). 9-1-1=20
> centers are required to have their clocks synchronized within 0.5 seconds=
=20
> of UTC. Kerberos/Active Directory requires clocks to be synchronized=20
> within 5 minutes of each other.
>=20
> If your log files have a resolution of 1 second=2C you probably won't see=
=20
> much benefit of sub-second clock precision or accuracy. If you are=20
> conducting distributed measurements with sub-microsecond resolution=2C yo=
u
> probably will want something more.
>=20
>=20
>=20
=
