he-lans...

vlan tagging and 802.1q

802.1q introduced the concept of the native vlan. the native vlan passes untagged frames
and are recognized as belonging to it as such.  other vlans are tagged, thereby
identified, via a two byte tag identifier or TPID which always has a value of 0x8100
(802.1q tag) and another two bytes are used for Tag Control Information (TCI), bringing the total to 4
bytes.  the tci contains a 3 bit priority field (PCP, priority code point) to identify class of service
(COS) and a 12 bit VID (vlan identifier). the single bit remaining and sandwiched between
the pcp and vid is the canonical format identifier (allowing compatibility for ethernet and
token ring). this will be set to 0 for ethernet (if it is set to 1, token ring, you are
working on a network from the last century) I actually supported a token ring network in
the 90's and it was great--back then.

802.1q does not encapsulate the frame although some authors refer to it as such.  it adds
(injects) this two byte field between the source mac and ether type field.  encapsulation
of frames for vlan identification is the purview of that other thing (isl) that cisco
should finally make go away and stop mentioning in cisco press books.

DTP as well as VTP are technologies that also should go away... Are you going to trust your
trunks to discover each other... no... are you going to deploy a new switch without
configuring the proper vlans... no... of course, no... will you save bandwidth and retain
your sanity... yes

even mentioning these things, you can almost see the authors cringing;

    "On critical trunk links in a network, manually configuring the trunking mode on both    
    ends is best so that the link never can be negotiated to any other state."

when is a trunk not critical... good Lord... it's embarrassing...