Layer 2 (Data Link Layer)

-: Welcome to Layer 2 of the OSI model, the data link layer. In the data link layer, we're going to package up the bits we got from Layer 1 and put those into frames and then, we're going to take those frames and transmit them throughout the network while performing some error detection, correction, identifying unique network devices using MAC addresses, and we're going to provide some flow control. Now, a MAC address is a media access control address which is a means for identifying a device physically and allowing it to operate on a logical topology. So, when we started talking about physical topologies in the last lesson in the last layer, we dealt with things physically, but now, we have to deal with things on a logical level. These MAC addresses are incredibly important for dealing with switches and other Layer 2 devices. When it comes to identifying MAC addresses, every manufacturer of a network card assigns a unique 48-bit physical addressing system to every network interface card they produce. As you can see here, we have 12-digit hexadecimal numbers that are used to represent these MAC addresses. These MAC addresses are always written hexadecimally wherein each of the letters or numbers is considered four bits. The first 24 bits or the six letters as you can see here identifies the particular vendor who made that card. In our example, we have D2:51:F1 and this is going to uniquely identify whichever person made this card. I like to think about this like a social security number in the United States. If you look at the first three digits, it's going to identify the state and the year that person was born in. For instance, if my social security number was 123456789, the first three digits might say when and where I was born. Maybe it was California in the year 1955 and those other six digits are going to uniquely identify me. Well, this is the same thing that happens with a MAC address. The first half of the MAC address, the first six digits, is going to tell us who made it. Was it made by Apple, Dell, Raw Link, or whatever? The second half is going to represent the exact machine it belongs to. This is important for our logical topology because we can look at the MAC address and observe the flow of data going through our networks. And at this point, we don't really care how these devices are physically connected. The issue at that point is a Level 1 issue. But now at Layer 2, we care about who's turn it is to talk and transmit so other devices aren't talking over each other. For example, when I teach this course in a classroom environment, instead of all of the students shouting out their answers at once, we use the system of raising our hands. We wait for the teacher to call on one of the students and then, we can let them ask a question. This is how we control the information flow so that everyone can hear each other. In a network, we use electronic mechanisms to do this same thing. Now, logical link control is going to provide connection services and allow your recipients to acknowledge the messages have actually gotten where you thought they were going. So, for example, if I called up and I asked if you got my phone call, you could say yes and that would acknowledge the receipt of that and then, we can move on to the next message. Logical link control does this for our networks. And because of this, it's the most basic form of flow control. Essentially, it's going to limit the amount of data that a sender can send at once and allow the receiver to keep from being overwhelmed. So, if I go back to my classroom example, if I'm sitting there and I'm moving too quickly, a student might raise their hand and say, "Hey Jason, I don't understand this. "Can you slow down and repeat it?" In the case of this video, you can just pause or go back and watch that part again, but in a classroom, they can't so they may ask me to repeat it. Logical link control, it similarly does the same thing, allowing a device to make this request for either less information at a time or to replay that information. Logical link control also gives us some basic error control functions such as allowing the receiver to inform the sender if their data frame wasn't received or if it was received corrupted and it does this by using a checksum. Now, since everything it receives is just a series of ones and zeroes, the receiver is going to add all of these up and the last bit will either be even or odd. If it matches, they add them all up and they're even, then, it's going to assume that this was good if you have received a zero, meaning it was even. If the last bit was odd, meaning it was a one, and they added up all the numbers and they got an odd number, that means it was good, as well. But if not, they can figure that something was bad and then ask for a retransmission of the frame. Now, communication can be synchronized across Layer 2 according to three different schemes. We have something known as isochronous mode which happens when the networks use a common reference clock similar to synchronous yet they also create time slots for transmissions, much like we did with time division multiplexing. This has less overhead than either of the other two modes because both devices know when they can communicate and for exactly how long. The second method we can use is known as synchronous method and this is much like we use back in Layer 1. It's going to involve devices using the same clock. But the reason it's different from isochronous is that this is going to allow us to have beginning and ending frames and special control characters to tell us when we're going to start and when we're going to end based on those beats. For example, if I use it in music, I have songs that have various time signatures, things like 3/4 or 4/4 timing. This tells us how many beats are in each measure. Our networks operate much the same way in that our devices can only communicate at frequencies specified by these particular clock cycles. Because of this, there isn't a lot of gap time that isn't already properly utilized and this becomes a major drawback for synchronized mode. And finally, of course, we have asynchronous which is going to allow each of our network devices to reference their own clock cycles and use their own start and stop bits. In this way, there's no real control over when the devices are allowed to communicate, though, and that becomes the major drawback here. Now, when we look at Layer 2 devices, we have things like network interface cards, bridges, and switches. In contrast to how a hub is a dumb machine that simply relies on a message coming in and repeating it back out, switches are smarter. They can actually use logic to learn which physical ports are attached to which devices based on their MAC addresses. And in this way, they can send data to specific devices in the network, allowing us to pick up and choose different lines of communication to go to different areas. Now, we'll talk all about how this works and how these switches do these, including things like CAM tables using the MAC addresses and how they're doing the switching across the network in later lessons and we'll go into depth in that because you will need to understand that to understand how networks really work. But for right now, just remember that switches, bridges, and MAC addresses are three great examples of things that operate at Layer 2, the data link layer.