Hello and welcome. My name is Tyler McMinn with Aruba Networks, and this is the Aruba Networking essentials series of videos. We are beginning part two with a focus on switching. In the previous part one set of videos, we went over a number of basic protocols, talked about the different layers that we can use to describe networking, and showed a very basic hands-on lab where we built a reachable network through a single switch using Aruba Cx software. In this set of videos, our focus is going to be on switching, going through our different VLANs and applying those to our Edge ports, configuring Trunking between two switches. We are actually going to introduce a second switch, which we started to in the very last video, but now we're going to be building it out entirely. Looking at loops with redundant links and how to solve that with Spanning Tree, another protocol, and applying link aggregation using a protocol known as Link Aggregation Control Protocol, or 802.3ad. Without further ado, let's get started. In normal traffic terms, when you're sending frames of information or packets of information, there's this risk of collision similar to when you're having a conversation with somebody and you go to speak, they go to speak, suddenly you're both speaking the same time and you can't really understand what they're saying. What nice people do is they will hear that the other person has spoken at the same time. They'll detect the collision, back off, and say, "No, you go ahead, I'll wait my turn." Well, the same concept applies in networking with the use of a hub or a repeater. This was one of the first devices that was used in networking and originally it was a repeater to extend the distances involved between your cable connections, where a network interface card would be listening to this link and if station B sends at the same time that station A, such a frame and a collision was detected, then, just like nice people, the network interface cards would detect the collision, back off, and simply try again in order to resend. We have a little animation there. This also occurs in the world of wireless, where in today's network with wireless, we have devices on the same wireless channel. When one NIC card or radio goes to the center frame and another one detects it they will again back off and try again. The protocols that handle this on the wired side is known as Carrier Sense Multiple Access Collision Detection. On the wireless side, it's the same carrier sense multiple access, however, with wireless, we use a collision avoidance mechanism similar to the collision detection. It just handles the wireless a bit better than the wired side. Humans, when they are in the same network or the same subnet, the same local area network, you can have a lot of devices or a lot of users in the same room all talking and they would end up having multiple conversations overlapping with each other; not really allowing for good communication. A lot of backoffs, a lot of having to re-listen for what's going on. A solution in the real world is you put these conversations in separate rooms. In one room, you have a meeting of people that are all listening and waiting their turn to talk, and in another room you have a different set of folks that are all listening and waiting for their turn to speak. Again, still one person at a time can speak, but you've divided your collisions into two different collision domains. How does this apply in the world of wired networks? Same sort of problem where we have too many devices that are all associated. If you could plug them each into their own individual hubs that were not connected to each other, then we could split up our collision domains without fear of causing a problem or causing collisions amongst each other. Use more hubs and split up the domain. Between these hubs, in the old days, we would use something called a bridge, which was the early version of a switch. It was just software driven rather than hardware driven. You would have separate connections for each of these hubs going down to ports on a bridge and the bridge would break up those collision domains between each other while still allowing communication amongst all your devices. In the world of wireless, we do the same thing, too many hosts in one collision domain, meaning they're sharing the same channel. What do you do? You get multiple access points, and with different access points, you've now got separate collision domains where only hosts and channel one are going to have to back off from each other while hosts in channel six are on a completely different channel, so therefore, they're not going to collide with channel one. In fact, we can lay these access points right on top of each other and there would be zero chance for a collision there. That's collision domains. In the world of switching, each interface is actually its own collision domain. Switches are smart enough to use through the use of MAC addresses toward each MAC address has a port that it's learned on and you would only send frames out as switch would only forward frames out that particular interface to that particular port or that particular MAC address destination. In the world of switching, every port is its own unique collision domain in that design. A broadcast domain is this idea of host C wanting to send a frame to everybody, to a broadcast destination. In previous videos, we saw an example of this through the use of our address resolution protocol. That would send to a destination MAC address at layer two of all Fs, which is the hexadecimal equivalent of all ones. It would also send at the same layer three, where layer two would be all ones, or all Fs. Layer three would be the address of 255, 255, 255, 255. All 255s means in binary it's all ones. Again, that's a reserved set of addresses at layer two, and at layer two that are designed to be broadcasted. What this switch will do when it receives a broadcast frame, is it will flood out all connected interfaces, except the one that came in on, so that it can reach all destinations within your broadcast domain. This is a great service add in IPV4, we adjusted slightly with IPV6 with something called an anycast address. In IPV4, what we typically use today, is these broadcast domains to allow flooding and allows applications like App to be able to do their job. Switches for broadcast, all foreign ports except the ones they received it on. The one they receive it on, we just call the ingress. The problem with broadcast is, you don't want them to flood out the rest of the world. If A is sending a broadcast, it's going to flood out within and all connected interfaces, by the layer two switch. However, your router in this scenario is smart enough to block those from flooding into the entire internet, or throughout your entire corporate structure. A routing device will not forward broadcasts. They essentially define the edge of a broadcast domain. This idea of flooding of collision domains and a broadcast domains is imperative to the basics of our switching mechanics. We'll come back to this a bit more as we continue to build out our switching labs. Quick pop quiz here, question number one, which of the following options below accurately describes collision domains, and broadcast domains? Take a moment, pause the video, give these are read-through. I'll show you the right answer, and then I'll show you why the other ones are wrong, and there should only be one correct answer. Think you know which one it is? All right. Let's take a look here. The correct answer is, a routing device defines the edge of a broadcast domain, that is correct. A, a collision domain relates to layer two processes while broadcast domains are layer three concepts. Not really. A broadcast domain is still a layer 2 concepts, so that's not correct. A collision or are common in modern network switches. Collisions are actually completely gone in a properly run at network. Modern networks, we should have zero collisions with everything being switches. Hubs would cause collisions, still allow for collisions. You're layer two net cards would listen for those, run their Carrier Sense Multiple Access Collision Detection protocol, our application and essentially to resolve collisions, recover from them. With hubs being basically outdated, and legacy devices at this point, always all modern enterprises should be running regular switches. You're running switches in your house. There's really shouldn't be running cross hubs. C, a multi-layer switch eliminates the need for a collision domain. That is incorrect. A multi-layer switch will still generally can allow the separation of collision domains, a multi-layer switch is basically a switch that can also route traffic. It doesn't eliminate the need for collision domains, that can still be something that you might run into. Generally, it's the switching itself that resolves collision domains in your network there. Really a multilayer switch deals with broadcast domains, it separates your broadcast domains. It's the switching piece, not just the multilayer's piece. This one's a little bit vague. Although you could somewhat consider that correct, it just depends on how you read it, so not a great answer. C or e, sorry, a router defines the edge of a collision domain. It does, but really focuses on the broadcast domain piece. That's really what a router is designed to do. Then broadcast domains are mainly a problem when you use a hub device. Actually collision domains are mainly an issue when you're using a hub device. They don't really impact broadcasts. Hubs were for broadcast the same as any packet, and really it's not, hubs don't even view the packets or the frames. What they look at is just the ones and the zeros. They're just like the toaster oven of network devices. Hopefully a little bit of success with that. In the next video, we're going to go in and actually start working with gear. We have a lot of labs coming up in these series of videos in part two. Thank you very much for your time. I look forward to going over these in the next video.
