wilmaethqn44
Member
OP
- Joined:
- Jun 2024
- Posts:
- 135
- From:
- Cardiff, UK
Hi all, sorry for the silly question maybe but I ran iperf3 for a week against my Contabo VPS and made some graphs. Everyone talks about Gbps this and Gbps that but honestly? My jitter was all over the place during peak hours and my little game server felt laggy even when speedtest said 900+ Mbps.
The consistent runs (low stddev) felt SO much better for my players. Is this a known thing or am I doing something wrong?
I can share the CSV if anyone wants, its not pretty though.
frames, tables, still valid HTML
tallinnying
Member
Night Shift
- Joined:
- Aug 2024
- Posts:
- 268
- From:
- Tallinn, EE
Anyway I grabbed your csv and threw it at gnuplot at 3am idk why I do this
Jitter vs fps drop correlation r=0.83 on my end. Used my InterServer box same region.
bash
For h in {1..168}; do iperf3 -c $host -J --get-server-output; sleep 3600; done
The 3am runs were best obviously. Network stack half asleep same as me
builds at 3AM, sleeps at noon
YuriDavid
Member
- Joined:
- May 2024
- Posts:
- 174
- From:
- Kyiv, Ukraine
NOTE: validating OP findings
---
- Contabo: 847 Mbps avg, 34ms jitter worst case
- InterServer: 612 Mbps avg, 8ms jitter worst case
- winner: InterServer for real-time workloads
---
WARNING: peak speed is vanity metric for UDP-based services
---
- collected 3 weeks data from 6 providers
- published: https://github.com
- methodology: hourly iperf3, parsed with python, graphed with grafana
Anyone can submit results. PRs welcome.
indentation is not optional
