0

Unless I am missing something, because switches can be chained together, I am not sure why the network layer is needed. Theoretically, one can dedicate a frame to any MAC address, and assuming there is a path in the graph of connected switches that connect those two PCs, the frame should arrive at the other PC at some point.

Because as far as I understand, if one chain multiple switches in a topology such as PCs A and B which are connected to switch C, and PCs D and E which are connected to switch F. Switches C and F are connected, then PC A should be able to send frames to PC D.

The only thing I am unsure about is how the MAC address tables of the whiches will look in the described topology, as switch C will see two MAC addresses incoming from the same port. Will they override each other as max age for the record is reached?

What am I missing here?

Aviv Aviv
  • 109
  • 2
    Simply put, MAC addressing doesn't scale to anywhere near a global network. Each switch needs to learn every MAC address present on the network, regardless of where it is connected. With on-topic business-grade switches that is often limited to 64k or less. – Zac67 Apr 04 '23 at 06:05
  • The short, plain answer is that Layer 2 (switches, in a very simplified version) is for building LANs. Layer 3 (IP network layer and therefor routers and other devices with IP addresses) is for building connections between LANs.

    The idea that you could build the internet on a scale beyond a few buildings or larger than a campus is like imagining the ability to build a network of roads with only 2 lane roads and stop signs to provide global transit. Works great for a small town but once you grow past that point you need a larger level of complexity and segmentation.

     – FrameHowitzer Apr 04 '23 at 17:57

0 Answers0