What contributes to DS ES issues?
Bit errors that survive all the error-correction processes that are turned on currently. 1 ES is one second-long period during which a bit error occurred, which means the maximum value that can be seen in 24 hours is 86,400. A very quick, concentrated burst of noise might cause 20-30 CRC errors, but only 1 or 2 ES's.
If a bit-error occurs on the line, it might get corrected by the FEC process - in which it contributes to the FEC counter.
If it doesn't get fixed by the FEC process, or the FEC process is turned off, it might get corrected by retransmission - in which case it contributes to some of the G.INP counters.
If it doesn't get fixed by retransmission, or retransmission is turned off, then it will contribute to a CRC failure - in which case it contribute to both the CRC counter and the ES counter.
My last line (18 months ago) was good enough to not trigger DLM intervention, so stayed on fastpath. It would regularly see 600-700 ES's per day, without issue.
Adding interleaving would likely sort it (at the expense of speed and latency). Adding retransmission would sort it, at the expense of losing tiny amounts of bandwidth whenever an error occurs.