[c-nsp] Optical module transmit power

[Richard A Steenbergen]

On Thu, Apr 30, 2009 at 03:37:44PM +0100, Michael Robson wrote:
> We have a selection of ZR modules (XENPAK-10GB-ZR) in 6500s that we  
> are using to drive some links at 10Gbps and I have recently noticed  
> that all receive values (as reported via the sh int te x/y trans  
> command) are lower than what Cisco specify as the minimum allowed  
> values for those modules (i.e. the minimum quoted is -24dBm whereas we  
> are seeing value reported as low as -28.8 dBm), and also a link that  
> is receiving at -28.9dB and flapping.

This is normal, it is very common for optics to "work" a bit below their
min rx spec. Like how elevator cables don't snap and let you plummet to
your death as soon as you hit +1 lbs over the maximum weight specified,
there is a built in safety margin in the specs to make sure the min spec
is achieved under all circumstances. Of course you're taking your life
into your own hands if you run it like this, as someone walking past and
bumping or bending the SMF cable could easily add enough loss to kill
the circuit completely.

When you do cross that magic line where the signal is no longer strong 
enough to work, flapping constantly is a common behavior on Cisco 6500s 
with XENPAKs (particularly on WAN PHY optics, which flap on any kind of 
error). On most other platforms you'll just start to see CRC errors and 
then eventually lose the link. Of course DOM on these things has been 
known to be wrong too (both due to hardware and software issues), but 
the behavior you're describing is normal so this probably isn't the 
case.

> For these modules, none of them are transmitting at anything like
> their maximum of +4.0dBm (Cisco's figures for the maximum transmit
> power), they are in fact transmitting between +1.9dBm and +2.3dBm.
> What determines what they will transmit at i.e. is it simply that
> better manufactured ones achieve a transmit value closer to the
> +4.0dBm power level, or is there some sort for decision/negotiation
> that determines the transmit value at connection; if the the latter,
> how can these modules be convinced to transmit at a higher power
> value?

You can't "adjust" the transmit power (well not on these things, on some
of the latest newfangled optics you can but thats a completely different
discussion), what it does it what it does. The transmit power spec is
basically telling you to expect your optic to put out a signal level
somewhere in the range between specified min and max, so you can plan an
optical budget that is both strong enough to work properly, but not so
strong that it blows up your amps or rx optics on the other side.

-- 
Richard A Steenbergen <ras at e-gerbil.net>       http://www.e-gerbil.net/ras
GPG Key ID: 0xF8B12CBC (7535 7F59 8204 ED1F CC1C 53AF 4C41 5ECA F8B1 2CBC)