[AusNOG] QinQ on Telstra Wholesale EA Fibre Services

[Mark Smith]

On 13 July 2017 at 11:04, Radek Tkaczyk <r_tkaczyk at hotmail.com> wrote:
> Hi Guys,
>
>
> We are looking at deploying a Telstra Wholesale EA Fibre service, and want
> to setup some QinQ tagging on the fibre link. Have done them before with
> AAPT Wholesale, but not Telstra Wholesale.
>
>
> Does anyone know if this can be done with TW EA Fibre?

I would be asking Telstra directly. If they tell you it can, you buy
the service and then it can't, you can demand recompense from them
(and if they're not forthcoming with that information, well then they
don't really want your business ...).

You can't do that (and certainly shouldn't try to do that) if you rely
on non-authoritative advice from the AusNOG mailing list.


> Are there any gotchas
> to watch out for with them?
>

If I remember correctly, don't send IPv4 packets with all zeros
destination addresses - it triggers a debug mode in their NTUs that
stops them stripping VLAN tags that they should be. That nice layer
violation cost me and my employer and Telstra a lot of time to resolve
- why should IP addresses in the packets matter to a layer 2 service?
(It was all zeros because I left the tester's IP test traffic
addresses as their all zero defaults, and I left it that way because
it shouldn't have mattered ...)

All zeros destination address is still a valid IPv4 address per RFC1122.


"There is a class of hosts* that use non-standard broadcast
         address forms, substituting 0 for -1.  All hosts SHOULD
_________________________
*4.2BSD Unix and its derivatives, but not 4.3BSD.




Internet Engineering Task Force                                [Page 66]

RFC1122                      INTERNET LAYER                 October 1989


         recognize and accept any of these non-standard broadcast
         addresses as the destination address of an incoming datagram.
         A host MAY optionally have a configuration option to choose the
         0 or the -1 form of broadcast address, for each physical
         interface, but this option SHOULD default to the standard (-1)
         form."

and now you know why there is a configuration switch for this in Cisco
routers' configuration registers

I'm guessing the developers of those devices thought that 0.0.0.0 as a
destination address didn't mean anything and didn't bother checking.
This is an example of why doing tricky things is a bad idea - they're
unexpected, may not work all the time and always as you think, and can
waste lot of somebody else's time.

Regards,
Mark.