The above image seems to be courtesy of the Cisco Learning Network, so thank you Cisco Learning Network!
“Frame Tagging” is adding a VLAN ID tag to an Ethernet Frame when traversing a Trunk, depending on what encapsulation you are using it has different rules, and one important note is that when a Frame traverses an Access port it does not get tagged – Only on Trunk ports. Now if the Ethernet frame
When looking at the encapsulation on the last post we had 3 choices from which to choose:
SW1(config-if)#switchport trunk encap ?
dot1q Interface uses only 802.1q trunking encapsulation when trunking
isl Interface uses only ISL trunking encapsulation when trunking
negotiate Device will negotiate trunking encapsulation with peer on
So you have “dot1q”, “isl”, or can set the switch to negotiate and work use either depending on the other switches encapsulation type.
There are 2 things that are common between ISL and Dot1q:
- They are both trunking protocols
- They are both point-to-point protocols
That is where the similarities stop, and the differences come in:
- ISL is Cisco proprietary (so no multi-vendor switching environments)
- ISL adds both a header and a trailer onto each tagged from, also known by the term of “double tagging” an Ethernet frame
- ISL encapsulates every frame going across the trunk, which every switch has to encap / decap every “double tagged” frame, costing time / CPU / Bandwidth resources per switch that has to process an ISL frame!
- ISL does not understand “Native VLAN” on switches, hence encapsulation of every Ethernet Frame
- Total overhead of 30 bytes – 26-Byte Header and 4-byte Trailer
So there is about nothing to like about ISL, I cannot think of any reason to be using it, and in some study materials it is claimed to basically being phased out – But we must know the differences from its nemesis dot1q.
However try to memorize why it’s so horrible, and the overhead it brings to the table.
Speaking dot1q, or IEEE 802.1q, I’ll list what that has going for it:
- Non-Cisco proprietary (can be used in multi-vendor environments)
- Understands Native VLANs, so frames in the default VLAN are not tagged at all, and any frames that show up at the Trunk with no Tag are assumed to be in the Native VLAN and allowed across the Trunk (unless it is otherwise configured)
- Adds only 1 Header known as “single tagging” and embeds the tag into the Ethernet frame itself, so sometime called “Internal tagging”
**I am researching but not finding much immediately on the difference between ISL tagging and internal tagging, or ISL not “embedding” its tag on the frame. If I find some clarification on this I will expand upon that**
So ISL could be known as “Double tagging” and dot1q as “Single tagging” and both have a 4-byte header, but they are in different places. ISL’s is in it trailer while dot1q has it as the VLAN ID in the Header (No trailer for dot1q).
Again, if it is traversing an Access Port the switch knows that port is in 1 VLAN only so it is NOT tagged when traversing it, only tagged over Trunk ports.
I will bookmark this VLAN topic here, as next is a hefty discussion about Native VLAN’s (why we need them), and get the lab involved with changing them around!