Week 4: Day 014 - Ethernet Basics #3

Hello again! Today I will be continuing my look at Ethernet Basics. Last time I talked mostly about the 10BaseT (and its fiber-optic counterpart.) Now I'm gonna talk about Enhancing Ethernet Networks. Basically, when a network is set up, it can be extended in many ways. Let me summarize it for you.

First, we have the simple fact that sometimes one hub is not enough. Before I continue, I would like to say that people don't really use hubs that much anymore, but that doesn't detract from the fact that there are, so this is still important to cover. Now, when you have too many nodes for one hub, what you have to do is get a new one of course! But how will you connect the hubs together? Simple! You use an crossover cable or uplink port. Regardless, it's a worthy question to ask, why do people who don't need a second hub still use one? Well it's because if all nodes are connected to one hub, if the hub crashes or malfunctions, the network will as well! Continuing from before, what is an uplink port? Well aside from the fact that they connect hubs together, they do it using a "straight-through" cable. To peform the hub connection, people "daisy-chain" them, which is basically plugging the cable into one regular port and the other into an uplink port. But remember, don't connect more than 4 hubs together! Also, remember that messing up is normal, since it is kind of tricky. Hub makers give uplink ports all sorts of names, the textbook says "MDI-X" and "OUT" are pretty common ones. If you wanted to use uplink ports as just a regular one, some hubs actually have a little button switch for that. By the way, remember that you may only "daisy-chain" hubs, and you should take your time making the connections. One relief is that if you mess up, it will not break anything, it just won't work.

Next, there are "Crossover Cables" which are special twisted-pairs that can be used to connect hubs together. The difference between crossovers and regular ones, is that the crossover reverses the sending and receiving pairs on one end of the cable, as described by the textbook. Since they're reversed, the hub can hear each other. This is why two standards are needed for connecting UTP cables to RJ-45 jacks. As a note, you can also daisy-chain crossover cables. Funny enough, many high-end hubs don't even have uplink ports, so it's necessary to use a crossover. Interestingly you can also connect two computers with 10BaseT NICs using a crossover, and no hub. Be careful about confusing the uplink ports and crossover cables, and remember that messing up does not have any large consequences (luckily.)

Now think of this, what if your Ethernet cables aren't long enough to reach their destination? Well there are these things called "Bridges"! These bridges are like repeaters (or maybe a hub) and connect two segments of ethernet cable. And it does more that too. It also filters traffic, which means stopping data from crossing over to another network, or it forwards traffic and that means passing the traffic from one side of the bridge to the other. Another note, nowadays wireless bridges are used more than Ethernet bridges. So in summary, it's basically a repeater, but it also monitors traffic that goes through it.

Finally, let's talk about the stuff that really matters! You'll see that in this modern day, we use switches a lot more often than hubs. And now we will lightly touch on this subject. Basically the government had more intensive and demanding applications that needed quicker version of the hub, something that provided more bandwidth... this device was called a "Switch". The problem with hubs is that, only one data message can be on a wire at any one time. On top of this, the hub was basically just a repeater, so if there was a collision, it would send out a signal asking for a resend of data. This becomes a problem when you have a big network. Everything suddenly becomes slower, and that's the reason why. Thankfully, the switch arrived to the consumer market. The switch pretty much looks like a hub, because like a hub, all nodes are connected to it using Ethernet. However, in the inside, they are much different. The switch is not only a repeater like it's predecessor. What it does, is that it takes advantage of the existence of MAC Addresses and gives conversations between computers maximum bandwidth that the network can handle. In my classroom, there is obviously switches here, to connect the lab together. I've already seen it in action, and like the textbook says, it acts like a hub, passing off data. But the intelligent switch copies the source MAC Address of the sender and creates a list of all the MAC Addresses on the network called a "Source Address Table" (SAT -- no not the dumb nationwide test!) When the table is made, the switch does great stuff with that data. The switch basically acts as if it were a telephone operator, as the textbook cleverly analogizes. When a computer sends a frame to the switch, it creates an "on-the-fly" connection between both the sending and receiving computer. Every conversation on the network is independent and individual (which is amazing) and each one runs at 10 Mbps. Each and every one port on a switch has it's own "collision domain" and the switch also buffers incoming data/frames. What's amazing is that two nodes connected to the switch can send data and not come across any collisions! However, with half-duplex switches, collisions can still happen, and in that case the CSMA/CD rules apply. Luckily, network devs figured out how to run switches at full-duplex mode, so each device can send and receive data at the same time. With a full-duplex Ethernet NO collisions will occur. How amazing is that?! On top of that each node gets full bandwidth of the network. At full-duplex mode you need not worry of the daisy-chaining rules. Since you can connect switches however you'd like, sometimes redundant connections are created. Back in the day of their release, this caused the network to crash, due to frames being caught in loops. The Ethernet standards body adopted the STP (Spanning Tree Protocol) to end those problems, according to my textbook. STP-enabled switches also use frames called "Bridge Protocol Data Unit" (BPDU) to communicate knowing the distances between each other and to track any changes that happen on the network. These hubs and switches work great for long periods of time, but sometimes there are problems with them. To troubleshoot there are three categories: "Obvious Physical Damage", "Dead Ports", or "General Flakiness". There are patterns which will alert you of the issues. It could be that a device connected, is not connected to the network, or maybe that there is obvious damage to the hub or switch itself. Sometimes, if the link lights are not flashing, try a different port. Finally, look at the cables and make sure they are not damaged, because it may be the culprit. On that note, I have finished this long chapter of Ethernet Basics. But it doesn't end there! The next chapter is all about Modern Ethernet, haha. Now I am officially 1/6th through this book. Wow! Thanks for reading, and tschuss!