There are two things about STP that confuse me.

Consider the following network:

enter image description here

There is a designated port from S1 to S3. There is also one from S1 to S2, and then S2 to S3.

If S1 sends an unknown unicast, it will forward it to S3 twice: S1->S3 and S1->S2->S3. Doesn't this defeat the concept of STP?

If S2 wants to talk to S3, will it only send it via Trunk 2?

  • 2
    S3 will be blocking frames on that VLAN from S2.
    – Ron Maupin
    Commented Nov 2, 2023 at 17:34

2 Answers 2


You are thinking in terms of direct point to point links between spanning-tree capable switches.

That is the most common case in networks today, but it's not the only case to consider. In general an Ethernet network could have.

  • Multipoint links (e.g. 10base-2)
  • Repeaters/hubs
  • Non STP aware bridges/switches.

While there have obviously been many tweaks over the years STP was designed in 1990. This was a time when Ethernet normally ran on coax, twisted pair Ethernet was just starting to come onto the market and bridges/switches were expensive devices that you tried to use as few of as possible.

If you think in this context, it makes perfect sense why STP is defined in terms of blocking ports, rather than disabling links. There may be stuff between S2 and S3 that is not visible to STP and needs connectivity to the network.


S3 port F 0/2 is in the blocking state. So it won't receive frames from S2.

Traffic from S2 to S3 has to go through S1.

  • And S2 will learn that by never seeing traffic from S3. (other than S3's BPDU's)
    – Ricky
    Commented Nov 2, 2023 at 20:15

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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