Hi
Gerald & team,
I’d like to share a feature proposal and get feedback from the community.
GitLab issue: https://gitlab.com/wireshark/wireshark/-/work_items/21126
The idea is to introduce an integrated TCP graph that combines several key metrics into a single view:
- RTT
- Bytes in Flight
- Throughput (send rate)
- Goodput (delivery rate)
Wireshark already exposes all of these, but they are currently spread across multiple graphs, requiring manual correlation and prior expertise.
This proposal is based on a Network Congestion Model I’ve been using in real-world troubleshooting. By observing the relationships between these signals, it becomes much easier to:
- Identify whether a flow is bandwidth-, buffer-, or application-limited
- Distinguish different types of packet loss (buffer drop, QoS, random loss)
- Diagnose performance issues in a more deterministic way
The goal is not to introduce new metrics, but to improve how existing ones are visualized and correlated.
I’ve included examples and PCAPs in the issue, along with videos demonstrating how the model is applied in practice.
I’d really appreciate any feedback, especially on:
- Whether this aligns with Wireshark’s graphing/UI philosophy
- Feasibility of integrating these metrics into a unified view
- Any existing work or similar ideas I might have missed
Thanks for your time.
Best regards,
Taifeng Tan