[j-nsp] optical transceivers recommended "Rx power" levels and signal measurement intervals

[Saku Ytti]

On (2013-06-16 20:40 +0300), Martin T wrote:

Hi Martin,

> 1) As I understand, XFP vendors store the recommended "Rx power" levels in
> transceiver non-volatile memory during the manufacturing process and those
> are read by switch/router over I2C bus. Are those values usually reliable?

I've found them reliable. But it's very coarse metric, you know you're
getting enough light but it does not say anything about actually your
ability to find the symbols in the light.
It's like measuring quality of music by dB, hmmm, that's bad analogy,
that's what we actually do.

> 2) At least in case of DPC-R-40GE-SFP line-card the "show interfaces
> diagnostics optics" displays Rx average optical power for SFP's. What is
> the measurement interval? I mean is it 10s average? 60s average?

The SFP specification which document DDM is publicly and freely
downloadable, I didn't verify, but I cannot recall it making any
requirements on poll rates.
There are probably quite few things which affect this, the HW in the optic,
the microcontroller reading the HW and populating the EPROM, frequency of
your vendor linecard reading the data off the I2C.

So my guess is, answer is 'it depends'.

-- 
  ++ytti