802.2/802.3

netwicked

Hey everyone,

I had a question regarding the sub-layers of the data link layer.

Odom's book says the IEEE assigned other functions to the 802.2 committee. I am just not really sure what those functions are. Why do we have an LLC header?

For anyone with Odom's book, I am referencing P61 of ICND1.

I get that the Type field tells the receiving host which L3 protocol the packet goes to. A value of 0800 would hand it off to IP.

Here is the direct quote from Odom:

If the 802.3 Type/Length field has a value of less than hex 0600 (decimal 1536), the Type/Length field is used as a Length field for that frame, identifying the length of the entire Ethernet frame.

So does that mean if the Ethernet Frame is 1536 bytes long then a 802.2 header would get inserted. I'm not getting the point of this 802.2 header.

What is normally used on networks today?

Why do we even need the sub-layer if we are running an IP network?

bighornsheep

The original Ethernet II standard by DEC,Intel,Xerox didn't have a type field. When the 802.3 was created, they needed to support multiplexing of IP, IPX, Appletalk etc... but need to provide backward compatibility.

The answer was to introduce type field codes which are much larger than the typical length of an ethernet frame. So anything less than 0x0600 is recognized as a "length" and greater than that are all field codes. Thus 0x8000 is IP..

Think of the MAC sublayer and LLC sublayer has "handlers" or "interfaces" of the Ethernet layer 2 protocol... MAC is used for Ethernet to communicate with the lower layer (physical layer), LLC is used for Ethernet to communicate with network layer such as IP, IPX, etc...

netwicked

bighornsheep wrote: »

The answer was to introduce type field codes which are much larger than the typical length of an ethernet frame. So anything less than 0x0600 is recognized as a "length" and greater than that are all field codes. Thus 0x8000 is IP.

Hey thanks for your reply!

Okay, so if the value in the Type/Length field is 0x600 or less, the the Type/Length field is used as a Length field. Since the field is being used to represent length, an 802.2 LLC header is inserted to identify the Type. Is that correct?

And also, if the Type/Length field has a value of 0x800, then the frame is seen as containing an IP Packet. But since 0x800 is in the Type field, how does the receiving node know the Length of the packet now?

bighornsheep

Sorry, I mixed it around. The original Ethernet II used the field for Type, 802.3 uses the field for length, and added the LLC header to take care of type field codes.

As for how Ethernet II calculated length, I don't see another place for this value, so I'm guessing it knows the frame delimiting points by reading the preamble and start-frame-delimiter.

http://www.dcs.gla.ac.uk/~lewis/networkpages/m04s03EthernetFrame.htm