Thanks, David. That's the kind of thing that I'd like to be able to 
display...pictures telling 1000 words, etc.

Ultimately, we're talking about linux (CentOS 6/7) running on modern multi-core 
hardware vs similar specs under some existing VSAN clusters (ESX 6).

As I mentioned, it's intended as a stratum2 setup with a pair of logical hosts 
in each of multiple locations, served both via an anycast VIP from each site 
for "good enough" clients (potentially many thousands of them) and via a DNS 
pool for true ntpd-based clients that support quorum. The latter will likely 
number in the high hundreds or low thousands. The pool and anycast setups are 
intended to allow horizontal scaling of the service as clients from more sites 
leverage it.

There's little doubt that qualitatively, the hardware servers will give better 
results, but the folks that manage the OpEx want to be able to quantify the 
negative impact that being on VMs might introduce. i.e. is it worth taking on 
the incrementally more expensive cost of 20-some server chassis to get the 
performance/accuracy increase over a similar number of VMs.

Do you mind sharing how you plotted the linked graphs?

Dan

Dan Geist dan(@)polter.net
(33.942962,-84.312118)
http://www.polter.net

----- Original Message -----
From: "David J Taylor" <[email protected]>
To: [email protected]
Sent: Tuesday, October 6, 2015 2:20:20 AM
Subject: Re: [Pool] benchmarking performance of ntpd on a virtual platform.

Greetings, all. I've been challenged to show some evidence that ntpd running 
on a VM (as a stratum 2 server in a company-internal pool) really does 
perform differently than on bare metal. Can anyone point me to any 
non-anecdotal scenarios or tests that might help demonstrate the 
performance/accuracy differences? One thought I had was to make a small 
cluster of several VM servers and several bare metal ones and add all of 
them to some clients and simply observe deltas in ntpq stats. Does anyone 
know of other good ways to get some "apples-to-apples" comparison data?

Thanks
Dan

Dan Geist dan(@)polter.net
====================================================

Dan,

For what it's worth:

Windows-10 "bare metal" GPS/PPS synched:
  http://www.satsignal.eu/mrtg/kiruna_ntp_2.html

Windows-XP running in a VM on the above hardware:
  http://www.satsignal.eu/mrtg/web-server_ntp-b.html

I don't have a comparison for Linux, and you don't specify what OS you wish 
to use.  What sort of accuracy do you need?

Cheers,
David
-- 
SatSignal Software - Quality software written to your requirements
Web: http://www.satsignal.eu
Email: [email protected] 

_______________________________________________
pool mailing list
[email protected]
http://lists.ntp.org/listinfo/pool
_______________________________________________
pool mailing list
[email protected]
http://lists.ntp.org/listinfo/pool

Reply via email to