Spanning Tree

aljugan

Hellow friends
i have one question about the spanning tree...as it mentioned in cisco book(ccna-icnd2..page 62 figure 2-1) that for example we have 3 switches and they are connected ....
my question is this this that when the packet is received by the destination switch-computer after that why loop does not ends because the packet is reached to his destination..for first time its ok this one is mentioned in icnd 1 book how mac table is created learning and flooding.......but i just want to know this that when a packet is flooded in the network it goes to his destination as well so after that loop should be finished....
Thanks

TheNewITGuy

The loop is from the switch's perspective

iamme4eva

In a stable network where MAC tables are fully populated, then unicast traffic would go directly to the destination.

However, broadcast traffic, ARP request, etc all broadcast out of all ports. This means that the frame exists in multiple locations, and will go round and round the network.

pamccabe

iamme4eva wrote: »

In a stable network where MAC tables are fully populated, then unicast traffic would go directly to the destination.

However, broadcast traffic, ARP request, etc all broadcast out of all ports. This means that the frame exists in multiple locations, and will go round and round the network.

Exactly. Spanning-tree works to prevent broadcast storms, MAC table instability, and multiple frame transmission.

NetworkVeteran

i just want to know this that when a packet is flooded in the network it goes to his destination as well so after that loop should be finished....

The crux of the problem is that the switches have no way of knowing the frame reached the destination. The very fact that the frame's flooded implies the destination is a broadcast, multicast, or unknown unicast address. The loop's presence is exacerbated by the fact that Ethernet frames do not have TTLs, unlike MPLS or IP packets.

blatini

I also have a quick question on this topic, but a little unrelated.

Can you manually configure the root switch, or is that always determined by the Bridge ID -> MAC address deduction?

Also a lot of diagrams I've been viewing with STP have the Designated Ports always on the root switch. Which makes sense. Is there a scenario where the designated ports wouldn't be there?

Any help is appreciated!

iamme4eva

A bridge ID is a priority + MAC address (for example 32768.aaaa.aaaa.aaaa, where 32768 is the priority).

So, what you should do is set the switch which you desire to be the root to a lower priority than the other switches.

EDIT:

Oops, I missed the second half of your question.

Yes, every port on a root bridge is a designated port. It doesn't need a root port, as it doesn't need a best path to get to the root...it is the root! And if there was a blocked link, the other end would do the blocking, as the other end would have a higher (or worse) priority. Every single segment has a designated port...on a segment that gets blocked, one end is a blocked port and the other is designated. The end that gets the designated port is the end with the lower priority, which would always be the root bridge...that's why it was elected in the first place!

NetworkVeteran

blatini wrote: »

Can you manually configure the root switch, or is that always determined by the Bridge ID -> MAC address deduction?

As imme4eva says, you can choose the root by setting the priority, and often you want to do so.

Also a lot of diagrams I've been viewing with STP have the Designated Ports always on the root switch. Which makes sense. Is there a scenario where the designated ports wouldn't be there?

Just about every switch in your network should have designated ports! At this point, you'd be well-served by either reading a chapter on spanning tree or attempting to setup a simple lab topology with a few hosts.

MickQ

For security reasons, you could program the desired root bridge to have a priority of 0.
Assuming that no other switches will have a lower MAC address and same priority, it will win out.

There's also BPDU guard and root guard which you'll learn about later

iamme4eva

NetworkVeteran wrote: »

Just about every switch in your network should have designated ports!

Ah, I misread that and didn't realise he meant designated ONLY on the root!

blatini

NetworkVeteran wrote: »

As imme4eva says, you can choose the root by setting the priority, and often you want to do so.


Just about every switch in your network should have designated ports! At this point, you'd be well-served by either reading a chapter on spanning tree or attempting to setup a simple lab topology with a few hosts.

Well, yeah... I am aware each switch has one port that is set to forwarding. I was getting confused in the terminology. It appears designated ports are on the root switch and the non-root switches each have one root port. Or maybe it is the other way around. Just different ways of saying the same thing it seems...

NetworkVeteran

blatini wrote: »

Well, yeah... I am aware each switch has one port that is set to forwarding. I was getting confused in the terminology. It appears designated ports are on the root switch and the non-root switches each have one root port. Or maybe it is the other way around. Just different ways of saying the same thing it seems...

The root switch has no RPs. All non-root switches have exactly one RP. The root switch's ports are all DPs. Most non-root switches have many DPs. The role names are critical because they're part of how STP ensures a loop-free spanning-tree.

theodoxa

blatini wrote: »

Can you manually configure the root switch, or is that always determined by the Bridge ID -> MAC address deduction?

Sort of. There is a command:

spanning-tree root primary

But, all it does is calculate and set the appropriate priority to ensure that the switch it is issued on becomes the root bridge.