Measure packet latency / drop between ingress and egress interface of a rate-limiting pass-through host

Question

Somewhere between my sender and receiver, there is host explicitly limiting my transfer rate.

I know it's a Linux host. The ingress interface for my traffic is eth0 and the egress interface is eth1. The rate limiter is most probably realized somehow using Linux' traffic control (tc), but I don't know the details.

From an earlier offline comparison between pcap files recorded at eth0 and eth1 interfaces, I know that the rate limiter adds significant packet latency (up to 15 seconds!) and packet loss (up to 16%), when I try to push through as much data as possible, using iperf.

All traffic is TCP/IP. Source IP, destination IP and destination port are known and constant. Only the source port changes between consecutive flows.

Given (root) access, is there a way to do a live measurement of the packet latency and drop counters / rates? Or recorded to a file as time series data?

Answer 1

Yes, you could. At first, you should watch your tc statistics output (see https://unix.stackexchange.com/questions/96804/tc-show-output-explanation)
Further, you could watch your network traffic flow with any network load utilities - iftop, iptraf,nethogs,bandwhich, etc.

P.S. And I'd recommend to review tc configuration of tc - it's better to use shaping than hard rate-limit. A good difference explanation is here: https://www.cisco.com/c/en/us/support/docs/quality-of-service-qos/qos-policing/19645-policevsshape.html