Spanning-Tree question

pham0329 · May 2011

Consider the following topology

A colleague of mine asked me if a host on Sw-A needs to access a host on Sw-D, what path would it take if Sw-E was the root?

From my perspective, looking at the topology, there's not enough information to accurately answer that question (ignore the orange light!)

My thought process on this question is as follows:

Sw-E is the root and all its port are DP
Sw-A’s link to Sw-E is its RP and remains Forwarding
Sw-C’s link to Sw-E is its RP and remains Forwarding

From that point, there’s not enough information to answer the question.
When Sw-D determines its RP, it looks at the lowest cost route. If both link have the same cost, it looks at the BID of Sw-A and Sw-C. If that is equal as well, Sw-D uses the port number as the tie breaker. So if the connection to Sw-A is on Fa 0/1 and Sw-C is on Fa 0/2, it will choose Fa 0/1 as its RP.

As such, unless we know what the BID, cost and interface of all the switch, I don’t think you can accurately answer that question. Is that correct or am I missing something? My friend insisted that all traffic will pass through the root, Sw-E...which, I don't think is correct. If Sw-D selects Sw-A as its RP, it's going to be blocking its link to Sw-C. So if all traffic goes through the root, when a host on Sw-A tries to access a host on Sw-D, it's going take the the following path

A > E > C > D, but if Sw-D is blocking its link to Sw-C because Sw-A is its RP, hosts on Sw-A won't be able to communicate with hosts on Sw-D

hiddenknight821 · May 2011

pham0329 wrote: »

Consider the following topology

A colleague of mine asked me if a host on Sw-A needs to access a host on Sw-D, what path would it take if Sw-E was the root?

From my perspective, looking at the topology, there's not enough information to accurately answer that question (ignore the orange light!)

My thought process on this question is as follows:

Sw-E is the root and all its port are DP
Sw-A’s link to Sw-E is its RP and remains Forwarding
Sw-C’s link to Sw-E is its RP and remains Forwarding

From that point, there’s not enough information to answer the question.
When Sw-D determines its RP, it looks at the lowest cost route. If both link have the same cost, it looks at the BID of Sw-A and Sw-C. If that is equal as well, Sw-D uses the port number as the tie breaker. So if the connection to Sw-A is on Fa 0/1 and Sw-C is on Fa 0/2, it will choose Fa 0/1 as its RP.

As such, unless we know what the BID, cost and interface of all the switch, I don’t think you can accurately answer that question. Is that correct or am I missing something? My friend insisted that all traffic will pass through the root, Sw-E...which, I don't think is correct. If Sw-D selects Sw-A as its RP, it's going to be blocking its link to Sw-C. So if all traffic goes through the root, when a host on Sw-A tries to access a host on Sw-D, it's going take the the following path

A > E > C > D, but if Sw-D is blocking its link to Sw-C because Sw-A is its RP, hosts on Sw-A won't be able to communicate with hosts on Sw-D

To be quite frank with you, this just made my brain cringed when I read this, because I recently studied and practiced STP, but it has been a month since I touched this stuff.

Don't confuse STP with frame forwarding logic. Your colleague was wrong to believe that STP would affect how the frames forward around the LAN. This is what the MAC address tables are for. Switch D needs Host D's MAC address and Host A's MAC address in its MAC table as well as Switch A. Remember, each MAC address in the table has an interface assigned to it, so that the switch knows which interface to forward out the incoming frames.

To clarify your initial question, I think your colleague may be asking how the STP frames circulate in the LAN. It starts with the Root Bridge. The Root Bridge always have no root ports. All their ports are designated ports. However, all non-root bridges would always have at least one root port. Once the STP frames arrive at the root port of the next bridges, the current switches then forward out the new Hello BPDUs with their own BIDs out the designated ports. Any ports that are blocked won't acknowledge or "update" the BPDU, but doesn't necessary mean they will ignore non-STP frames as well. Here's how the STP works in the topology you gave us, which is sufficient enough to figure out as long as SW_E is the root port, and all path costs are the same.

EDIT: [ignore the diagram below, since it didn't come out neatly. I lost spaces when I posted it.]
SW_E ---> SW_A ---> SW_B
---> SW_D
--->SW_C ---> SW_B (will not acknowledge)
---> SW_D (will not acknowledge)

Hopes I finally help uncringe that brain of yours as well

Forsaken_GA · May 2011

hiddenknight821 wrote: »

Don't confuse STP with frame forwarding logic. Your colleague was wrong to believe that STP would affect how the frames forward around the LAN.

The entire point of STP is to create a loop free topology. STP most certainly effects how frames are forwarded around the LAN. Try turning STP off and then creating a looped network. Let me know how quickly the broadcasts make the network unusable.

To answer the OP, all traffic doesn't necessarily pass through the root. It all depends on which links go into blocking, and which don't. This is why placement of the root bridge in your topology is a very important choice.

instant000 · May 2011

to put it another way, for all intents and purposes, this is your topology:

Switch E is the root bridge, and notice how B and D have basically equal paths to take (no information is available to identify the better paths to take)

Without knowing path costs, bridge ID, interface, there is nothing to say which interface would
be placed in blocked or forwarding mode, as this diagram is missing information.

Look at my spin on the diagram, hahahaha, Notepad rules!

                         [Switch B] 
                      _____|     |_______
                 -----                    -----
    ------------                              --------
     |                                                |
[switch A]---------------[Switch E]-----------------[Switch C]
     |                                                |
      ----_____                               ______---
               _______                  ______
                      ------|     |------             
                           [Switch D]

Forsaken_GA · May 2011

Yup, one link on switch B is going to block, one link on switch D is going to block, as that's where the loops can form.

If the orange marks on the original diagram indicate the links that are blocking, then switch A will send traffic directly to switch D, as that would be it's only functional path. If the link between A and D blocks instead, then the traffic goes E-C-D, because that is the only valid path.

pham0329 · May 2011

hiddenknight821 wrote: »

To be quite frank with you, this just made my brain cringed when I read this, because I recently studied and practiced STP, but it has been a month since I touched this stuff.

Don't confuse STP with frame forwarding logic. Your colleague was wrong to believe that STP would affect how the frames forward around the LAN. This is what the MAC address tables are for. Switch D needs Host D's MAC address and Host A's MAC address in its MAC table as well as Switch A. Remember, each MAC address in the table has an interface assigned to it, so that the switch knows which interface to forward out the incoming frames.

To clarify your initial question, I think your colleague may be asking how the STP frames circulate in the LAN. It starts with the Root Bridge. The Root Bridge always have no root ports. All their ports are designated ports. However, all non-root bridges would always have at least one root port. Once the STP frames arrive at the root port of the next bridges, the current switches then forward out the new Hello BPDUs with their own BIDs out the designated ports. Any ports that are blocked won't acknowledge or "update" the BPDU, but doesn't necessary mean they will ignore non-STP frames as well. Here's how the STP works in the topology you gave us, which is sufficient enough to figure out as long as SW_E is the root port, and all path costs are the same.

SW_E ---> SW_A ---> SW_B
---> SW_D
--->SW_C ---> SW_B (will not acknowledge)
---> SW_D (will not acknowledge)

Hopes I finally help uncringe that brain of yours as well

I disagree with your first paragraph. STP has a direct affect on the path a frame with take around the network. If a link is blocked by STP, frames are certainly not getting forwarded out that interface.

I don't think he was talking about the Hello BPDUs, I think he was talking about host-to-host communication.

pham0329 · May 2011

instant000 wrote: »
to put it another way, for all intents and purposes, this is your topology:

Switch E is the root bridge, and notice how B and D have basically equal paths to take (no information is available to identify the better paths to take)

Without knowing path costs, bridge ID, interface, there is nothing to say which interface would
be placed in blocked or forwarding mode, as this diagram is missing information.

Look at my spin on the diagram, hahahaha, Notepad rules!
                         [Switch B] 
                      _____|     |_______
                 -----                    -----
    ------------                              --------
     |                                                |
[switch A]---------------[Switch E]-----------------[Switch C]
     |                                                |
      ----_____                               ______---
               _______                  ______
                      ------|     |------             
                           [Switch D]

Haha...I tried to diagram the topology using notepad but gave up after 10 minutes and switched to packet tracer. That diagram does make a lot more sense than my original diagram

Thanks for confirming that without additional information, you won't be able to tell which port on Switch D and B will be the RP or gets blocked.

hiddenknight821 · May 2011

pham0329 wrote: »

From my perspective, looking at the topology, there's not enough information to accurately answer that question (ignore the orange light!)

I actually took the liberty to make an instance of your topology in Packet Tracer to test to see how the double loop works with one root bridge, and I failed to read your instruction carefully. So, to answer your question is you were right. There were not enough information.

Forsaken_GA wrote: »

The entire point of STP is to create a loop free topology. STP most certainly effects how frames are forwarded around the LAN. Try turning STP off and then creating a looped network. Let me know how quickly the broadcasts make the network unusable.

Thanks for pointing that out. I realized I was wrong about how STP may have affected the frame forwarding now.

His topology has me fooled. I definitely need to review STP before I take the exam way after I start WGU. This is coming from someone who hasn't been in school since the end of March, and doesn't lab everyday. Thanks.

instant000 · May 2011

pham0329 wrote: »

Haha...I tried to diagram the topology using notepad but gave up after 10 minutes and switched to packet tracer. That diagram does make a lot more sense than my original diagram

Thanks for confirming that without additional information, you won't be able to tell which port on Switch D and B will be the RP or gets blocked.

If I could have interpreted the amber ports as a link problem (or blocked ports), then those links would not get used, and I would have known the path that would get created.

Once you said to "ignore the orange lights" that's when I didn't have enough information.

Spanning-Tree question

Comments