Trunks - is the packet re-encapsulated between switches?

[andersl]

Hi,

I'm working as a Cisco instructor, and one of my students asked me this question: If a packet have to pass by multiple swithces, is the packet ISL or dot1q encapsulated only one time (the frame is only tagged at the entry-switch, and untagged at the end-switch) or will each switch untag on incoming interface, and re-tag on the outgoing interface? I'm not able to find the answer to this one when I search CCO...

Please also answer by email (anders@a-team.no) if possible!

Thanks in advance!

-Anders

[edehaas]

anders,

It's my understanding it is encapulated or tagged once. I would not swear by that however. If you find the "real answer" will you post it?

Thanks.

[bsivasub]

The header is stripped off and encapsulated again if it needs to go another trunk port.

This is for multiple reasons,

1) Packet may not need to trunked again

2) Packet may have to go through a different encapsulation like ISL if the initial encapsulation is dot1q and vice versa

and there may be other reasons

The only exception to the rule that i am aware of is dot1q tunneling

http://www.cisco.com/univercd/cc/td/doc/product/lan/cat6000/sw_6_3/confg_gd/dot1qtnl.htm#xtocid236921

[patoil]

I have a similar question. Assuming static membership:

Is the frame encapsulated/tagged since it enters an access port? or VLAN ID is only valid over trunk links?

I mean, the broadcast domain inside the switch is defined by VLAN identification information carried by each frame, or the switch as a whole entity suled by an O.S. can decide which are possible destination ports based on then source interface VLAN ID?

(Instructors are often punished with this kind of procedural questions)

[bsivasub]

There is no tag for access port but the switch does use its internal tag for the vlan in which the source port belongs but it is not a dot1q or isl headers type of tag. Switch uses its intelligence once the packets enters a port to switch it accordingly.