[c-nsp] NTP Servers

[Erik Soosalu]

There was discussion about NIC HW timestamping on the NTP mailing list
recently.

I didn't read the whole thing, but one of the issues that was brought up
was "How accurate is the clock on the nic?".

For very high precision, you'd have to discipline the NIC clock as well,
so then you get twice the issues.


Thanks,
Erik 

-----Original Message-----
From: cisco-nsp-bounces at puck.nether.net
[mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Saku Ytti
Sent: Monday, June 25, 2012 11:24 AM
To: cisco-nsp at puck.nether.net
Subject: Re: [c-nsp] NTP Servers

On (2012-06-25 10:50 -0400), Josh Baird wrote:

> And guess what the Infoblox appliances run?  :)

Earlier poster also wondered why waste money on "NTP appliance'. Some of
these appliances have hardware timestamping, which will significantly
increase accuracy, if your network is low-jitter and low-delay (like
most
HW switched networks today are).

Curiously at least on ingress side your random NIC can support HW
timestamping today, and IIRC even egress. Couldn't be arsed to surf
intel.com through the datasheets.
I wonder if the NIC HW timestamping function is flexible enough to
inject
timestamp in NTP packets as very first thing in ingress and very last
thing
as egress. And if there is way to give the NIC accurate timing somehow.

Probably quickly becomes cheaper to buy some appliance than try to
figure
out how to hack this in NIC driver, possibly kernel and NTPd.
-- 
  ++ytti
_______________________________________________
cisco-nsp mailing list  cisco-nsp at puck.nether.net
https://puck.nether.net/mailman/listinfo/cisco-nsp
archive at http://puck.nether.net/pipermail/cisco-nsp/