Gigafast first hop latency

[iahmed]

Hello

I tried to raise this issue some time ago but I wasn't able to get an answer. Latency is lower on 500 Mbps but higher on 900 Mbps.

The first hop to 90.247.64.1 is 10ms which is a considerable amount.

# mtr --report --no-dns 8.8.8.8
Start: 2022-09-20T22:48:42+0100
HOST: mbp                         Loss%   Snt   Last   Avg  Best  Wrst StDev
  1.|-- 172.17.0.1                 0.0%    10    0.5   0.6   0.5   0.7   0.1
  2.|-- 90.247.64.1                0.0%    10   10.0  10.6  10.0  11.1   0.3
  3.|-- 63.130.104.194             0.0%    10   11.2  10.8  10.2  11.2   0.3
  4.|-- 72.14.216.236              0.0%    10   11.8  11.7  11.5  11.9   0.1
  5.|-- 216.239.41.193             0.0%    10   14.2  13.8  13.3  14.2   0.3
  6.|-- 142.251.54.27              0.0%    10   12.1  12.5  12.1  13.1   0.2
  7.|-- 8.8.8.8                    0.0%    10   12.5  12.4  12.1  12.7   0.2

The latency between London and Frankfurt is ~2ms so the physical distance can't be the issue.

Can you please escalate this to a network engineer?

[CrimsonLiar]

10ms (1/100 sec) is a little high, but it's not totally excessive.  I'm curious though about the early routing there, it does look a little odd.  Are you using the standard THG3000 ?

[iahmed]

I'm using a unifi router on 172.17.0.0/22. (Yes the issue is present on default router).

With regards to how high, it depends on how you look at it. For the average user, it may seem insignificant but it isn't. It can affect many types of applications such as gaming. My workload is database related and latency has a serious impact on write queries. A few statements can push the total time into hundreds of milliseconds or even longer.

Latency is reduced if the package is downgraded so perhaps the network engineers can start there.

[Jayach]

---

@iahmed wrote:

Latency is reduced if the package is downgraded so perhaps the network engineers can start there.

---

Good luck finding a network engineer in Vodafone.

[Ripshod]

For some of us this has been an ongoing problem for over 3 months with no sign of a resolution. And we're just the ones who know what's happening. 

[CrimsonLiar]

So what is the reason for using the 172.16.0.0./12 address range.  Needing 1M+ local addresses is quite uncommon!

[iahmed]

It’s /22 which gives 1022 useable addresses.