0

I see one Course Lecture as follows:

enter image description here

I trouble in understanding "source address for the next node’s reply", What is the meaning of next node's reply?

3
  • neither do i. all headers at all layers have source addresses and destination addresses. how exactly these are used depends on protocol involved, but usually they both are used at some point. But it is highly protocol-dependant on where and how. If layer 2 is reliable and sends ACKs (e.g. WLAN) it uses source address to know where to send ACK to. However, since most communication patterns involve some kind of bi-directional communication (e.g., if it was a TCP packet, an ACK will be sent), it is highly likely that if B has received a frame from A, then B will send a frame to A shortly after.
    – Effie
    Oct 20 at 10:51
  • This way, it makes sence that B learns A's MAC address from A's frame, i.e, it caches the address.
    – Effie
    Oct 20 at 10:53
  • this statment has very little sence unless it specifies concrete layer 2 (and possibly layer 3) protocol involved, so you probably have to ask the author what he meant.
    – Effie
    Oct 20 at 11:08
2

This is a poorly worded (IMO) description of the fact that the receiver will (likely) need to respond to the sender in some way: either to reply to the message or at least acknowledge the receipt. At the very least, the receiver will need to distinguish between the sources of several received messages.

1

Indeed, on wired Ethernet layer 2 there's no notion of connection nor error message sent back to the sender, so there's no reply expected.

The fact is the Ethernet standard do mandate that the source mac address be sent in the frame.

This allow two important things:

  • the receiving device can cache the mac address in its arp table, avoiding an unnecessary ARP lookup. This is probably the point the author wanted to make, since it is very likely that the content of the frame will receive a reply.
  • the network bridge (switch) can update its mac-address table, associating the mac address of the sending host with the port on which the frame ingressed

For example if a web browser request a web page, this is TCP traffic that will be encapsulated in a IP packet, which will in turn be encapsulated in multiple Ethernet frames.

The frames themselves don't expect any kind of response, but the TCP datagram do expect an answer that will come back, likely, by the same path, and when this answer comes, the router will already know the Mac address of the original requestor.

As pointed out by @Effie some layer 2 (wireless) may have a notion of reply, but this will be specific.

11
  • layer 2 can have notion of connection or layer message, it depends on layer 2. Most wireless layer 2 have notions of ACKs.
    – Effie
    Oct 20 at 10:55
  • Exactly - the source MAC address is used for MAC bridge learning and little else. "Avoiding an unnecessary ARP lookup" might create security problems, so I don't think this is widely used.
    – Zac67
    Oct 20 at 11:49
  • Thanks , bad habit to think mainly about wired Ethernet. I slightly edited my answer but feel free to provide an additional answer (or edit mine if you wish)
    – JFL
    Oct 20 at 11:50
  • I disagree. There is an (perhaps unstated) assumption that 2-way communication is taking place, so the receiver needs to know who to reply to. Also, the receiver needs to know who sent the message so it can distinguish between several received messages.
    – Ron Trunk
    Oct 20 at 12:09
  • yes, but usually, receiver needs to know "identity" of the sender, which is not the mac address, of the last sending hop. And filling ARP cache is an optimization, it is not necessary.
    – Effie
    Oct 20 at 13:16
1

From the perspective of the data link layer/L2, a frame in the data link layer needs [...] a source address for the next node's reply is nonsense.

There is no such concept in the data link layer. A valid source (MAC) address may be required in many L2 protocols, but it doesn't do anything with it.

The next upper layer - usually the network-layer protocol, or potentially an application-layer protocol riding directly on L2 - might use the L2 source address. That L2 address might even by required by the upper layer, but that isn't anything the data link layer itself has to worry about.

1
  • Ethernet may do nothing with it, but other layer-2's may. And the various layer-"3" protocols that don't have any addressing of their own use it. (ARP, for example.) As Ron points out, it's poorly worded, but the src is usually required so a reply can be made, not that one necessarily will be.
    – Ricky
    Oct 20 at 23:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.