[AusNOG] TX/RX Conventions

[Lincoln Dale]

On Wed, Jun 13, 2012 at 1:45 PM, <john at netniche.com.au> wrote:

> I'm not aware of any convention other than the "flip it if it doesn't work
> and curse shutdown-by-default policies".
>
> Cisco appear to place TX on the "left" of the optics (first port). The LC
> connector specification chickens out from making a distinction, and the
> SFP/SFP+ specifications also don't make a commitment (they allow for copper
> and single-fibre, after all).
>

There is a standard here and every transceiver/optics vendor follows it.
Its not explicitly spelt out but if you really wanted to know, take a look
in SFF-8431 table 3 or figure 1 and you have the electrical contacts used
on SFP/SFP+ and where TX and RX are physically.
The practicalities is that no optics vendor is going to 'crossover' the
optical side of rx and tx relative to where the electrical connections are
and put a more expensive PCB with more layers inside to do so.

SFP/SFP+ transceivers are _meant_ to have an arrows embossed on the metal
jacket/sleeve showing the transmit/receive directions. (i say meant to
because not all do, but I took a photo of one that does, see
https://arista.box.com/s/720c5c52227da1199658 and you can see the arrows)

On Wed, Jun 13, 2012 at 3:06 PM, Tom Sykes <TomSykes at nbnco.com.au> wrote:

> When you've got access to both pieces of equipment and you check status
> lights then I agree - "swapping" the fibres until it works is probably
> acceptable. But when the fibre technician doesn't have access to the
> equipment (for example: it's 10km away in some other place with a number of
> patches in the middle), it helps to agree on a convention.
>

You guys are probably dealing with lowest-common-denominator in 'fibre
techs' here - BUT - the chances are that said techs do have appropriate
technology for detecting what pair of a fiber has light and someone posted
it already.  Its fair to say that just about everyone would have a
smartphone with a camera on it, and said camera's sensor will happily show
you light at 850nm or 1310nm.

may be harder for you to write such a thing into a spec, but this is one
nice thing about camera cmos sensors and no filter blocking said parts of
the spectrum.


cheers,

lincoln.


Lincoln Dale | Principal Engineer, Arista Networks Inc. |
ltd at aristanetworks.com
au did: +61 3 9999 7442 | m: +61 417 457 965 | us did: +1 408 547 5782
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ausnog.net/pipermail/ausnog/attachments/20120613/6137a342/attachment.html>