Network Engineering Stack Exchange is a question and answer site for network engineers. It only takes a minute to sign up.
Sign up to join this community
What's the type of multiplexing used by Ethernet (LLC layer)?
It says in Wikipedia "The LLC sublayer provides multiplexing mechanisms that make it possible for several network protocols (e.g. IP, IPX and DECnet)". I don't understand this multiplexing principle. What do they mean by "multiplexing"? I can't find the information.
Thanks
Multiplexing is the use of a single communication channel (medium, resource) for multiple, mostly independent purposes. An example from the analog world is a traffic light that could be considered to use time-division multiplexing (TDM) to enable the use of a street crossing for vehicles going in intersecting directions. There are many other forms of multiplexing.
Ethernet frames carry a payload. For multiple types of payload (IPv4, IPv6, IPX, ARP, BPDUs, AppleTalk, PPPoE, ...), their type needs to be specified in a universal way (rather than looking at the payload and trying to interpret it). That way enables some kind of deflector to redirect each frame received by a NIC to the service that handles that type. Without multiplexing, you'd need a dedicated NIC and network for each type of traffic.
With Ethernet, the EtherType field is used for that purpose or, even more universally, a SNAP header.
For example, an IPv4 instance sets the EtherType value for its frames to 0x0800. It passes those frames to a NIC to transmission. On the destination, the frames are received and their EtherType value indicates that they need to be passed to the IPv4 instance there, enabling the IPv4 stacks to talk to each other, without exclusive access to Ethernet.
The same happens with IPv6 (EtherType 0x86DD), PPPoE (EtherTypes 0x8863 and 0x8864), or any of the other protocols that can run on top of Ethernet (or another protocol using the same mechanism).
On the other hand, imagine a mail service that transport envelopes between two commercial buildings. As long as there's no writing on the envelopes, the sender may differ but the receiver is always the same. That receiver would have to open the envelope and figure out where to pass the letter on. That might require a lot of guesswork and if the letter is in a foreign language, they'd be lost.
Instead, the sender marks each envelope with a number/name/whatever that unambiguously identifies the intended company in the other building, where the mail person can easily identify the purpose and destination of the letter.
Of course, using a fully qualified address on a letter envelope in real life enables us to multiplex a single instance of mail service between arbitrary senders and recipients worldwide. Perhaps you could compare the EtherType to writing the destination country on the envelope, enabling delivery even when the actual address is in Chinese.
