Should I expect random (or even regularly-occurring) bit errors to cause corrupted/dropped PDUs (packets) proportionate to the PDU size ?
Troubleshooting high (0.01% to 0.1%) packet loss over a short (<1ms) 2-hop path, 1Gbps+.
I ran these two pings concurrently.
Pretty sure their combined traffic volume shouldn't be enough to cause (too much) contention between the two.
Would probably do well to gather a larger sample size / longer run, but here we are for now.
ping -c 1200001 -s 4 -W 2 -qni 0.001 10.0.0.1 --- 10.0.0.1 ping statistics --- 1200001 packets transmitted, 1199002 received, 0.0832499% packet loss, time 1222942ms ping -c 1200002 -s 1192 -W 2 -qni 0.001 10.0.0.1 --- 10.0.0.1 ping statistics --- 1200002 packets transmitted, 1199224 received, 0.0648332% packet loss, time 1235676ms
Wouldn't we expect (over a long enough sampling period) for the 1220 byte packets to be dropped proportionately more often than the 32 byte packets ?
Yet above, the 32 byte packets see higher loss ...
Can we reasonably infer, then, that the cause of the loss is likely not bit-error-related ?
Bonus : Someone advised me that ...
32 byte packets are far more lossy then 1220, because of all the excessive headers, and CRCs and so forth.
Is this true ?
What is the mechanism behind smaller packets being inherently more lossy ?