[c-nsp] IEEE 802.1ad vs Cisco's Q-in-Q

[Tassos Chatzithomaoglou]

Here is another strange thing:

In CatOS:

http://www.cisco.com/en/US/products/hw/switches/ps708/products_configuration_guide_chapter09186a00801dd67e.html#wp999435

"When a tunnel port receives tagged customer traffic from an 802.1Q trunk port, it does not strip 
the received 802.1Q tag from the frame header; instead, the tunnel port leaves the 802.1Q tag 
intact, adds a 1-byte EtherType field (0x8100) and a 1-byte length field, and puts the received 
customer traffic into the VLAN to which the tunnel port is assigned. This EtherType 0x8100 traffic, 
with the received 802.1Q tag intact, is called tunnel traffic."


In IOS:

http://www.cisco.com/en/US/products/hw/switches/ps708/products_configuration_guide_chapter09186a00800da6fc.html#wp1013415

"When a tunnel port receives tagged customer traffic from an 802.1Q trunk port, it does not strip 
the received 802.1Q tag from the frame header; instead, the tunnel port leaves the 802.1Q tag 
intact, adds a 2-byte Ethertype field (0x8100) followed by a 2-byte field containing the priority 
(CoS) and the VLAN. This Ethertype 0x8100 traffic, with the received 802.1Q tag intact, is called 
tunnel traffic."


Am i missing something here (8100 hex = 1 byte?); Are these 2 totally different implementations?

Tassos


sthaug at nethelp.no wrote on 14/11/2006 11:45 μμ:
>> I think 802.1ad defines 0x88a8 as the prefered outer tag (different
>> from 0x8100 of 802.1q), which makes it much easier for switches to
>> know whether this is a single or a double-tagged frame.  And i think
>> Extreme & Juniper follow this.
> 
> Extreme uses 0x88a8 as the default outer Ethertype, but newer switches
> let you configure any value, including 0x8100.
> 
> Juniper uses 0x8100 as the default outer Ethertype.
> 
>> On the other hand, Cisco uses what i would call a "hack", since it
>> seems to support double-tagging by just using the single-tagging
>> code (checking if the ethertype is 8100).
> 
> The Cisco solution may be a hack, but it's a very *useful* hack. Given
> our operational experience, we would probably not use 0x88a8 for the
> outer Ethertype even if all equipment supported it - it's simply not
> flexible enough.
> 
> It *might* be useful to have a different Ethertype than 0x8100 for the
> outer tag if you could define this on a per-port basis. For all switches
> we've looked at so far, however, it is defined on a per-switch basis.
> 
>> IMHO, the best solution would be to be able to configure both
>> inner/outer ethertypes manually.
> 
> I agree that it is useful to be able to configure the outer Ethertype.
> I don't see the use for configuring the inner Ethertype.
> 
> Steinar Haug, Nethelp consulting, sthaug at nethelp.no