136 IPExpert Advance Spanning Tree Features BackboneFast

[Music] backbone fast is another Cisco proprietary extension to the spanning tree protocol the purpose of the backbone fast is to allow switches to find an alternative path to the route should they lose their route port you don't act for backbone fast to work it needs to be enabled globally on the switch which means it cannot be an able enabled only on some interfaces or some villains it's a global operation on the switch furthermore in order for it to be effective and fully operational in our network the backbone fast must be enabled on all the switches in the network the reason for that is because of what happens once the backbone fast is triggered I'll talk about it in few seconds what triggers the backbone fast well what triggers the backbone fast is the loss of BPD used on the route port or not only that but what happens if we receive a superior BP new on one of the ports that is blocking if we receive a superior BP new on the blocking port that means that we should no longer be blocking on this port if this is indeed a better path to reach the route but we cannot do that immediately in traditional spanning tree because we have to wait for the message to expire and that adds twenty seconds to the complete convergence to the full convergence in our network backbone fast aims to bypass the max range timer but in order to do that it really needs to determine whether the new path through the blocking port is indeed the best path to reach the route instead of the current route port that we have in order to determine that the switch needs to interactive interactively figure that out and it is going to do that using another cisco proprietary protocol called the root link query the root link query is a protocol that looks very similar to spanning tree if you would to look at it at the frame level the format is just about the same however the address to which it is and is not that of the spanning tree in a case that we have some switches that don't understand the root link queries we wouldn't want to confuse them so if they receive these frames they are simply going to ignore them and the switch that generated the root clink queries is going to have to go through the standard spanning tree convergence instead of the slightly fast one using the back bound fast there are two components of the root link query there is the request message and the response message and as you are used to by now I'm going to show them in action before I jump into the terminal and start showing you BPD frames and bits and bytes in their headers let's take a look at the example in which the backbone fast can be of use and speed up the conversions in our network for that I'm going to use the network of three switches they're going to be my cat one cat two and cat three cute one will be the route in this network and my switches are going to be interconnected in the triangle that looks like this this is going to be fast me 24 and this is going to be fast in at 24 this will be faster than 0 20 and this will be faster than 0 20 and this one here will be 22 given the network like this with all equal cost links and with cat 1 being the root the converged spanning tree topology will be that these two ports are designated ports that this becomes the root port this becomes the root port this becomes the designated port and this one here will be the blocking port which means we are not going to be sending any DPD frames on this port here now what would happen to the traffic on let's say hosts behind cat 2 here shoot the link to route fail so if this link here fails what's going to happen is that cat 2 will immediately transition to become a root what would it become a root well because it's no longer receiving VPD frames from cat one and because this here is a blocking port and by definition it doesn't send any bpdu frames get to the house things it is the root bridge which means that this here remains the designated port but now cat to start sending the frames that from cat trees perspective here are inferior why are these frames inferior well these frames are inferior because it is still receiving frames from the good route from the route that was already elected in our network and if cat one was elected route in the network consisting of cat1 cat2 and cat three with all links being up it is still going to become a route it is still going to remain a route from cats trees perspective even though this link between cat1 cat2 is down so when these frames start arriving to cat three caps tree is not going to accept them immediately it's not going to do anything about them for the duration of the max-age timer the max age by default is 20 seconds once these 20 seconds expired well it's not exactly 20 seconds I should I should make it clear that there is going to be the time between the previous frame being received and the 20 seconds but it's going to be around 20 seconds but it's not 20 seconds exactly this that this timer is max age - the lifetime of the frame but those are really semantics so let's say that after 20 seconds fast internet 20 here is going to start transitioning from the blocking phase into a listening phase here we are going to listen for the incoming VP news at this point here we are going to start forwarding the frames from the good root out because when we transition to the listening phase we are going to start forwarding the beep news when these VP news arrived - cat - it is going to stop sending these frames here and it's going to realize that there is a better route in our network but this port here Fastnet 2003 is still in a listening phase and it's going to remain there for 15 seconds after that we are going to move into the learning phase where we are going to remain for 15 seconds and only after that we are going to move into the forwarding state where this port here becomes the designated port and this port here becomes the root but all during this time until this here happens until the sport transitions into the forwarding state these hosts behind cat 1 host a B and C are effectively cut off from the network for how long well for 20 seconds plus 30 seconds here that means that for around 50 seconds these hosts cannot communicate with the rest of the network this is how the traditional spanning tree would converge let's see that in action so I have the configuration here on my cat one cat 2003 that matches the diagram that I was just explaining so if I do show spanning tree and cat 1 here I will see that I have two ports and they are both designated ports and they're both forwarding in cat - if I do show spanning tree I will see that I have two ports one is designated the port facing catalyst tree and fastly 24 the one facing cat 1 is the root port and is falling on cat 3 as expected I will have two ports of which one will be blocking the one facing kit - let's now go ahead and shut the link down between cat 1 and cat two before I do that what I'm going to do is I'm going to go to cat 3 and I'm going to say debug spanning tree events what I want to see with this debug is what happens when I shut that port down also on cat - I'm going to prepare the show spanning tree output I will have to be rather quick with this I have to do this show command within 15 seconds or sorry 20 and after shutting down the port so I'm going to go to cat one I'm going to shut the port down and I'm going to wait on cat two for the interface to go down and then I'm going to do show spanning three I will go back to that output but what I want to show you here is that now on cat 3 I'm receiving information about the new route if I go back to cat - I see that this bridge is the route which means that this port festival at 20 is now designated port however the port on cat 3 has only now just before I switched to this terminal changed topology changed the state into a listening phase and we can see here that when that happened we sent a topology change now we can see here that fast internet 20 is now listening it transitioned into the learning phase and only now it has transitioned into forwarding state that's me that means that since I've done shut down until this moment here before this highlighted line here happened there was no communication between hosts that may have been behind cat 2 and the rest of the network because the link between cat - and cat 3 was being blocked on get 3 side it was being blocked until the max age timer expired when the max age expired then while the port was transitioning through the listening and learning phases of the spanning tree traffic was still not flowing so how does backbone fast improve on this for that you're going to go back to the white board let's redraw our topology so here I have get one I have death doom and cap 3 I have the links between them this one here is the route this is fast net 24 fast may 24 last night 20 same here and fast in at 22 on this side here now we know that this here is the blocking link that this one here is the designated port this one is the root port designated designated and the root port let's again examine what happens when this link here fails cat two is going to declare itself as the root it is going to start sending the inferior bpdu frames but now because we have enabled backbone fast on cat - on catch one and cat three when this frame arrives on this port here what cat three can do is it can send out the root link query to cat one out of its root port asking cat one hey are you the root and if cat one or is this correct is this the best path to reach the root one cat one receives this frame is going to respond back saying yes this is correct this is the best path to reach the root when this happens and this is going to happen very quickly this port here is immediately going to transition into a listening phase which means that we don't have to wait for max-age to expire in listening phase we are going to spend 15 seconds then we are going to go into the learning phase for 15 seconds and finally they're going to move into the forwarding state which means that it is still going to take about 30 seconds for this convergence to happen but we don't have to wait those additional 20 seconds let's see this in action I'm going to go back to my terminal where in the meantime I have undid all the changes that I have done before so now all the links are back and operational and if I do show spanning-tree on cat one I'm going to see the two interfaces on cat - I'm going to see the two interfaces and on cat 3 I'm going to see the two interfaces and as expected faster than at 20 the one facing cat - is in a blocking state I'm now going to go and enable spanning tree backbone fast on cat 3 I'm going to do the same thing on cat tube and I'm going to do the same thing on get one before I proceed to shut down the link between cat one and get to which I'm preparing right here I'm going to go to cat 3 and I'm going to enable the debug of spanning tree events and I'm also going to enable the debug of the backbone fast the command for that is debug spanning tree backbone fast so I'm going to go to cat one and I'm going to shut the link down and then immediately go to cat 3 and let's observe what happened there the first thing that I see here is the time hearing from the new route this is because cat do now thinks it's the route the next thing here that I see is that I have received and inferior bpdu on faster than at 20 that's the one facing cat - I'm sending the rlq request PDO towards cat 1 on faster than at 22 and I have received a response on fast internet 22 cat 1 confirmed that that indeed is the best path to reach the route and I can see here that immediately fast internet 20 moved into the listening phase I didn't have to wait for 20 seconds to expire before I moved into this phase of operation now when this happened my port facing cat to start sanding VP do frames now the forwarding of the production traffic of the data traffic from the stations that may have been connected behind cat to and care tree is still not flowing through this interface but cat 2 is now forwarding the BP news received from cat 1 which in other which had the result that cat to now received a superior BP new frame and it's generating the topology change here on fast it left 20 so I'm receiving the topology change notification generated by cat 2 as the result of receiving a superior BPD and then finally we're going to move into the learning phase we are going to change we are going to generate our own topology change notification because now we are transitioning into the forwarding state so we can see here that we have bypassed twenty seconds of waiting time before we have bypassed the twenty seconds of waiting time that was originally there as the result of the max age so this is what backbone first does but before I proceed and just end here because this is what the backbone fast is let me just show you this bit here in Wireshark I want to show you these this root link queries flowing through so I'm going to go ahead and I'm going to do no shutdown on the link between cat one and cat too and I'm going to prepare my monitoring on cat three I'm going to say monitoring session one source interface Fast Ethernet 22 I want to monitor traffic going in and out and I'm going to send a copy of this traffic to faster than zero three and I'm going to preserve any markings that I may have on these frames and here what I'm going to do is I'm quickly going to enable the filter that I don't want to see any loopback frames I don't want to see CDP and I don't want to see DTP the only thing that I want to see here are really other frames because I don't have any other traffic this will be the app this would be a spanning tree frames and these would be the frames that maybe something other than the protocols that I'm filtering out here so now I'm going to go back to cat one and just make sure that my spanning tree is now in a pristine state that means that all the links are operational so here on cat - I see the same thing and on cat 3 if I do show spanning tree I should be seeing the familiar situation that fast at 20 is blocking so now I'm going to go to cat 1 and I'm going to save shutdown before I do that I'm going to restart my capture here and I want to observe what is happening and I was actually very lucky because the very - the very first two frames that I'm seeing here are the root link queries so I can see here this is the actual query being sent out but because this is a very very proprietary protocol from Cisco Wireshark cannot really decode it all I'm getting here is just data it doesn't understand this protocol but we know what it does so we can kind of look backwards and understand what is going on so this was the query this here was the response to the query and as the result we can see that the topology change has actually happened in our network we have seen now how backbone fast works will it's configured properly in other words when it's configured in our entire network which is one of the prerequisites for the operation of the backbone fast but what would be the result what would be the behavior in our network should we turn off backbone fast on one of our switches like for example on cat one let's try that out so I'm going to go to cat one and I'm going to say no spanning tree backbone fast it is now turned off on cat one I can confirm that if I do show spanning tree back bound fast and I get the result that the backbone is turned off it is still on on cat - and on cat three so show spanning-tree backbone fast here on cat three shows me that the backbone fast is turned on let's now try to turn off the link between cat one and cat two and observe the result on cat 3 so here I am going to cut one interface fast let's 24 I'm going to prepare this link for shutdown but before I do that I want to confirm that my debug zone catch we are still running I'm running the debug for the spanning tree events and for the backbone flask so now I'm going to go to cat one and I'm going to say shutdown on our link let's see the result what I'm seeing here now are the hellos the BPD news coming from cat - and I can also see that cat - that tariqat 3 is sending the root link clearly requests only one one on fasted net 22 it is sending the ARL cues to cat one but cat one now cannot respond which means that we are going to be receiving these BPD use from cat talk until the max-age expires in other words configuring backbone fast on cat 3 and cat two did absolutely nothing in our network to improve the convergence style because we didn't turn it on on cat 1 and this is the reason why the backbone fast needs to be enabled in the entire network in order to be fully operational

Channel: CCIEORDIE.COM

Views: 1,257

Keywords:

Id: 374MjiogeGM

Length: 21min 24sec (1284 seconds)

Published: Thu Feb 08 2018