I want Advice on Configuring Multiple NTP Servers for Optimal Performance

Hello everyone,

I am in the process of setting up multiple NTP servers to contribute to the NTP Pool Project. My goal is to provide reliable and accurate time synchronization services across different geographic regions. I want to ensure I configure them in a way that not only optimizes their performance but also benefits the pool as efficiently as possible.

Is there a suggestion approach to distributing traffic load across multiple servers? Should I prioritize certain servers based on location or network capacity?
What are the best practices for securing NTP servers from abuse or potential attacks like amplification DDoS?
Are there specific tools or metrics I should focus on when monitoring the health and performance of my NTP servers??

Thank you… :grinning:

Hi and welcome to the pool.

Just to clarify are you looking to be in the production or distribution side of the business? I see production as having some clock source connected to your server (GPS based clock for example) and adding it to the pool where the distribution side is something like a VPS with a hosting company that is sync’ed to other clocks but still answers millions of requests from clients.

Either way if you are able to have the servers in different physical locations, close to the clients you want to serve then that is better. I have a number of hosted VPS servers, in the distribution side, serving requests across Asia mostly.

Distribution of traffic to your servers is done by the pool DNS servers handing out your servers ip address in response to requests for pool.ntp.org and others. When you add your server the pool decides what geographical area it is in and adds it to that zone. The pool DNS servers do geoDNS lookups and send out the servers ip address to clients in that area depending on the connection speed you have set for your server.
When you start to add your servers it is a good idea to start with a low connection speed and then increase it as you find out what your server can handle in the way of requests. For example my server in Sydney is set to 100Mbit and receives about 50 million requests a day while a similar server in Singapore is set to 3Mbit yet receives over 500 million requests per day.

As for what to monitor - I only have VPS systems so if you are in the production side of things others may have better suggestions but I have just added some metrics so I can keep an eye on them. The public page is NTP server stats

Have fun :slight_smile:

1 Like

The second link makes me suspect this is an unusually well-done bit of search engine optimization. It has nothing to do with NTP and lots to do with recruiting tech employees.

Mr. hssmithh goes to Greenwich?

2 Likes

You replicated the link via quoting the spammer’s post. You may want to edit your own post to obfuscate the link.