0

It's pretty easy to see that packets transferred via internet take not a constant amount of time, even if the two communication partners don't change the location and the route in between likely doesn't change:

$ ping pool.ntp.org
PING pool.ntp.org (129.70.132.37) 56(84) bytes of data.
64 bytes from stratum2-4.NTP.TechFak.NET (129.70.132.37): icmp_seq=1 ttl=50 time=29.0 ms
64 bytes from stratum2-4.NTP.TechFak.NET (129.70.132.37): icmp_seq=2 ttl=50 time=29.3 ms
64 bytes from stratum2-4.NTP.TechFak.NET (129.70.132.37): icmp_seq=3 ttl=50 time=36.0 ms
64 bytes from stratum2-4.NTP.TechFak.NET (129.70.132.37): icmp_seq=4 ttl=50 time=25.1 ms
64 bytes from stratum2-4.NTP.TechFak.NET (129.70.132.37): icmp_seq=5 ttl=50 time=26.9 ms
64 bytes from stratum2-4.NTP.TechFak.NET (129.70.132.37): icmp_seq=6 ttl=50 time=26.2 ms
64 bytes from stratum2-4.NTP.TechFak.NET (129.70.132.37): icmp_seq=7 ttl=50 time=32.0 ms

When I run this locally for ~5 minutes, I can see the minimum to be 29.0 ms and the maximum is 179.7 ms.

I can imagine the following components to influence the time:

  1. My local machine: I guess if I use my network otherwise a lot, the packet could be in a queue and it could take some time until it is read.
  2. Air from my machine to my router: I guess the radio waves send by my computer to the router travel faster / slower depending on (a) the temperature and (b) how many devices send something on the same frequency.
  3. My local router: We have ~10 devices connected to our local WLAN router. Similar to my machine, I guess packet are sequentially sent (or at least the amount of parallelism is limited).
  4. Routers in between: Similar to my local router
  5. Cables between my client and the pool.ntp.org: Similar to my local

But what are the biggest factors increasing the variance?

(Please note that I didn't ask where most time is spent - I guess this is simply (5))

closed as off-topic by Zac67, Ron Maupin Sep 24 '18 at 17:09

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "NE is a site for to ask and provide answers about professionally managed networks in a business environment. Your question falls outside the areas our community decided are on topic. Please visit the help center for more details. If you disagree with this closure, please ask on Network Engineering Meta." – Zac67, Ron Maupin
If this question can be reworded to fit the rules in the help center, please edit the question.

  • As has been frequently noted here, ping is not a precise tool to measure latency or RTT. Additionally, networks not under your control (the Internet) are off-topic here as the discussion would lead to guesswork or opinion-based answers, both of which are explicitly off-topic here. – Zac67 Sep 24 '18 at 16:25
  • Welcome. You might like a paper about research in this topic citeseerx.ist.psu.edu/viewdoc/… – jonathanjo Sep 24 '18 at 17:36
  • @Zac67: If this is not the right StackExchange site, then what is the right one? – Martin Thoma Sep 24 '18 at 19:29
1

In general variation is caused by queueing. Actual transmission delays tend to be relatively constant.


My local machine: I guess if I use my network otherwise a lot, the package could be in a queue and it could take some time until it is read.

Indeed.

Air from my machine to my router: I guess the radio waves send by my computer to the router travel faster / slower depending on (a) the temperature and (b) how many devices send something on the same frequency.

The actual travel time for the radio waves is negligable and the transmission time for something as small as a ping packet is also negligable.

However your computer may have to wait until the channel is clear before transmitting. If the channel is highly congested this may add noticable delay.

My local router: We have ~10 devices connected to our local WLAN router. Similar to my machine, I guess packages are sequentially sent (or at least the amount of parallelism is limited).

Right.

Furthermore if you are on a home or small buisness network the connection to your ISP may be over a shared medium, such as cable TV coax or GPON. So as well as waiting for packets in it's own queue the router may have to wait it's turn among other users of the shared medium.

Routers in between: Similar to my local router

Yep, more potential queuing in routers and switches.

Cables between my client and the pool.ntp.org: Similar to my local

Cables themselves are usually point to point nowadays, but there are likely to be devices such as Ethernet switches, ATM switches, MPLS routers, IP routers carrying tunneled packets etc that don't appear in a traceroute but nevertheless have queues.

Not the answer you're looking for? Browse other questions tagged or ask your own question.