Cisco CCNA 200-301 – EtherChannel Part 3

6. StackWise, VSS and vPC

This lecture you’ll learn about Cisco’s multishass Ether Channel options of stack wise VPC and VSS. Matching Ether Channel settings have to be configured on the switches on both sides of the link. You know that already. You can configure separate port channels from a switch to redundant upstream switches like you see in the diagram here. I’ve got my access layer, access Three switch and it has got a port channel with bundled links going up to CD One. And it’s got another separate port channel with other physical interfaces that are bundled into a port channel going up to CD Two.

The problem with this is that spanning three will see those separate port channels as two separate interfaces and it will block one path if a loop is formed like it is here. So in our example, CD One is the root bridge. So the port channel from access fee to that is going to be forwarding, but the port channel going up to CD Two would form a potential loop. So spanning three is going to block that port channel. So I’ve got four physical interfaces configured as uplinks here, but only two of them are forwarding.

So I’m only getting half of my possible bandwidth. So that brings us back to the same problem of only using half of our available physical bandwidth. Again, we configured that same topology in the lab in the last lecture. So things were better after we configured Ether Channel. This is how it was before we did it. So again, CD One is the root bridge and you see from access three it’s got those four physical uplinks, but only one of them was forwarding traffic, which was the interface connected to CD One and exactly the same on Access four as well.

So four physical parts, only one of them forwarding traffic. After we configured Ether Channel we have two out of the four forwarding traffic now rather than one. So it’s better. And actually I’ve still got the lab running, so let’s go on there and just verify that. So I’ll go on to Access three and do a show Ether Channel summary and you can see that I’ve got my two port channels. One is going up to the first core distribution where I switch, the second one is going up to the second verbal.

If I now do a show spanning three VLAN ten, which is the VLAN that my PCs are in, I can see that those two port channels, one of them is forwarding traffic. This is the port channel connected to the root bridge and the other one is blocking because it’s forming a potential loop. So still not perfect. If we go back to the slides, there is a way that we can get this configured so that we are actually forwarding on all of our physically connected uplinks. And the way that we do that is with multiclass Ether Channel. So some switches will support that when we configure it the switches support a shared Ether channel from different switches. So you see the diagram I’ve got here.

Now if I go back one slide or a couple of slides, normally we would have two separate port channels, one going up to CD One and another one going up to CD two. When those core distribution layer switches support multishass Ether channel, though, I can do a single port channel which goes to both of them. Now again, the switches must be configured with matching settings. When you configure Ether channel, you need matching settings on both sides of the links. So here, axis three has got a port channel going upstream and switches CD One and CD Two have got a shared port channel going downstream. So CD One and CD Two need to be advanced level switches that support talking to each other and having a shared port channel going downstream.

So that is supported on some models of Cisco switch. When you do configure this, spanning three is still enabled, but it only sees a single logical interface on both sides. It sees a port channel going upstream from access three and it sees a single port channel going downstream from CD One and shared with CD Two. So it sees that as just a single link. So there’s no loop there. So spanning three does not block any of the ports. We have all of them forwarding. We get the full use of all of our physical bandwidth. It supports full load balancing and redundancy across all the interfaces. If any of the interfaces goes down, it will fail over to another available interface. So the technologies that Cisco have available that support this are StackWise, VSS and VPC. Now, a single switch is not going to support all three of these. Different platforms of switches support different options.

And if a switch does support one, it will only be one. So the switch will either support Stack Wise or it will support VSS or it will support VPC, not two or three at the same time. And this is only on the higher end switches. If you’ve got one of Cisco’s lower end switches, then this is not going to be supported. So which option is supported on which family of switches? Well, Stack Wise is on Catalyst switch platforms, including the Catalyst 37, 50, 38, 59,000 notice I’m just seeing including here, because Cisco come out with new models of switches very regularly and when they do come out with a new model of switch, it’s likely to support one of these three options. So this will be updated over time. If you have a look at the picture here, this is a stack of switches that have been configured with StackWise. StackWise uses special proprietary cables at the back of the switches.

When you do configure a Stack wise Stack, the separate physical switches all operate as if they’re one switch and they are configured as if they’re one switch as well. That is similar to the next option which is VSS, the virtual switching system that’s supported on the Catalyst 4506 500 families. And the last option is one of it’s a little bit different is VPC virtual port channel that’s supported on the Nexus switches. And rather than stacking the switches together and then being managed as if they’re a single switch, when you use this on the Nexus they’re still configured as two separate switches but you do a matching configuration on them and that allows for a shared port channel from those switches.

Okay, so those are the MEC options at the CCNA exam level. Cisco expects you to know a broad overview of them that they don’t expect you to know how to configure these. If you go on to CCMP level or one of the other CCNA tracks like Data Center, you’ll learn then how to configure and monitor these technologies.

7. Layer 3 EtherChannel

You’ll learn about layer three Ether channels in the earlier lectures. In this section you saw how to configure layer two Ether channels. We can configure layer three Ether channels as well. You can see that with the configuration example here. And the configuration is actually exactly the same. The only difference is that we make the ports a layer three port with the no switch port command. So that’s a standard command you can use on a layer three switch whether you’re using Ether channel or not, to say that the port is a layer three port so that you can put an IP address on there. So let’s look at the configuration. So a global config. I’ve said Interface range gigabit Ethernet 10 one to two. And then I say no switch port to make those layer three ports. And then I’m going to configure them as a port channel. So I say channel group one mode and then the same options as we had for layer two. It can be Active, Auto, Desirable, on or Passive depending on whether you want to make it a static Ether channel with on. Or we can use Active or Passive for LACP or we can use Auto or Desirable for PAGP.

So that will group the ports together into an Ether channel. We can then put our other configuration on that Ether channel. That’s where we’re going to put our IP address. So we say Interface port channel one. The one ties up with a channel group number. And then in this example we’ve said IP address 1921-6812-5252-5252 and remember to do a no shutdown on the interface. Okay? So that is a layer three E for channel. That’s a complete config. So this could have been a really short lecture. But there’s another implication of this. Something else I want to talk to you about here, and that is the network design. So looking at our traditional campus design with the core distribution and access layer and the way that this would be done originally would be there would be layer two links between the axis layer switches and the distribution layer switches.

And the default gateway for the end hosts would be on the distribution layer switches and we would have spanning three running between the axis and the distribution layer because they’re layer two links. Well, a trend that you can see increasingly now is actually putting layer three links in everywhere. And the benefit you get from doing that is it means that you’re not using spanning tree anymore. Spanning tree is a necessary evil. The bad side of spanning tree is that it tends to shut down half of your links and it’s also very slow to recover from failure as well. So it’s bad, but it still does good things. You still need it because it would be way worse if you had a layer two loop in your network because that would just basically bring your network down.

So we make sure that we don’t have layer two loops by having spanning three enabled. But the bad side is that it’s slowly recovered from failures and it does shut down half our links. So it would be better if we could have layer three links everywhere and not use spanning three anymore. Right now you may be wondering, well okay, well why didn’t we always do that? Why did we used to have layer two links between the access layer and the distribution layer? Now in a lot of modern networks we still do have layer two links from the access to the distribution layer. But an increasing trend that you’ll see now is that layer three everywhere and the reason it’s possible is that layer three switches have come down in price a lot.

Back in the day, layer three switches were a lot more expensive than layer two switches and your access layer switches were where you had the high port count, where you had all of your hosts plugged in. So you had a lot of access layer switches and it just wasn’t possible from a cost point of view to put layer three switches in on all your access layer switches. But because time has moved on now and the prices have come down for layer three switches, it is more feasible to do that now. So if you do put in layer three switches everywhere, including in the axis layer now, you actually have the layer three links from the access layer to the distribution layer, meaning you’ve got an IP address on risk port here and an IP address on risk port here. So because everything is layer three, you don’t have spanning tree running anymore. Spanning three only works on layer two links.

So the way that you do have the paths being determined between your switches, it’s not being determined by spanning tree anymore, it’s being determined by your routing protocol. So when you’re using layer three links everywhere, you will configure a routing protocol on here as well and it’s up to the routing protocol to handle the path determination with your routing protocols, they support equal cost load balancing. So you’re not going to have links being shut down, all your links will still be available and traffic will be load balanced across all of them. Also, your routing protocols recover from failures, they converge a lot faster than spanning three days. So that’s the benefit you get there. Now, when you do, this last thing to tell you about it is that the default gateway is not going to be on the distribution layer switches.

Now, as it would be with layer two on the axis layer, the default gateway for your end host is going to be on the axis layer switches. And if you think about it, it has to be because we’re going to have an IP subnet down here and we’re going to have a different IP subnet here. And the default gateway for your hosts has to be in the same IP subnet. So your default gateway is now going to be configured on your access layer switches, and you’re going to have routing everywhere. Layer three links everywhere above that. Okay? And I put this in the Ether Channel lecture, because very often when you’re doing that, you are going to have efer channels going up between your switches. And when you’ve got layer three everywhere, of course going to be layer three Easter channels. Okay, that’s everything I needed to tell you here. See you in the next lecture.