CCNA Notes: How Ethernet And The Network Switch Came About

Any sufficiently advanced technology is indistinguishable from magic.

Arthur C. Clarke

In the early networks, computers were connected using coaxial cables (IEEE standards 10BASE5 and 10BASE2) in a bus topology. This involved one cable connecting many separate computers. Any data transmitted would be delivered to every device on the cable. Only one device could transmit at a time, and it could only transmit in one direction; this is known as half-duplex. If two devices transmitted data simultaneously, a collision would occur. This caused all devices on the cable to stop transmitting for a random amount of time, this prevented all devices from starting to transmit again at the same time, causing a loop. A message would also be sent to the rest of the network to indicate that a collision had occurred. To prevent collisions, devices on the network would try to sense if another device was transmitting data. This method was known as Carrier Sense Multiple Access with Collision Detection (CSMA/CD).

Move forward a few years, and Twisted Pair cables were introduced. These are the Ethernet cables used in most of today’s networks. They originally consisted of 2 pairs of wires twisted together. They are twisted to mitigate the effect of electromagnetic interference (EMI). When released, this standard was known as 10BASE-T. 10 meaning megabits per second, and -T meaning twisted pair.

The way these cables work is by sending data down two wires (TX), and receiving data on the other two (RC).

By this point, the networks of the future were starting to take shape, networks were moving to the star topology we know and love today. A network device was introduced called a hub, these were Layer 1 devices. Hubs looked bit like switches and had a similar function, but unlike switches, any data they received would be sent out of all ports. Furthermore, like the networks before them, they only operated at half duplex.

But the introduction of hubs created a problem. If a PC is connected to a hub, the PC will send data down 2 wires of the Ethernet cable, pins 1 and 2 in the RJ45 header. It will then receive data on pins 3 and 6. The hub will do the opposite, so it will send data on pins 3 and 6, and receive data on pins 1 and 2. But what would happen if you plugged a PC into another PC? Or a hub into another hub? It simply would not work. These cables are known as straight-through cables, because both sets of wires go to the same pin outs on the other end of the cable. So pin 1 on one side, went to pin 1 on the other, pin 2 went to pin 2, and so on.

Because of this, a new type of Ethernet cable needed to be created. This was called the crossover cable. It meant if two hubs were connected, they could both send data down pins 3 and 6, and receive data on pins 1 and 2. This was possible by making pins 1 and 2 on one end, go to pins 3 and 6 on the other.

Any device that transmits data on pins 1 and 2 and receives it on pins 3 and 6 is known as an Medium-dependent Interface (MDI) device. Any device that transmits data on pins 3 and 6 and receives it on pins 1 and 2 is known as an Medium-dependent Interface – Crossover (MDI-X) device.

But hubs just could not stop causing problems. Now if we connect two hubs together, only one device between the two hubs will be able to communicate at once, and if more than one communicated at once, they would collide. Because there was more devices, more collisions would happen, and more down time would result from it. It was therefore necessary to create new ‘collision domains’. A collision domain is simply an area of a network where a collision can occur. It doesn’t matter how many hubs you connect together, there would only be 1 collision domain.

This prompted the introduction of bridges. A Bridge was a layer 2 device and brought the introduction of MAC addresses with them. Bridges got much closer to what a switch does now, it’s job was to forward layer 2 traffic based on the destination MAC address. Bridges could connect multiple hubs together to create new collision domains. It meant that if there was two hubs joined together by a bridge, the devices on one hub would not be affected by communications from devices on the other hub. If two hubs are joined together with a bridge, this creates two collision domains. The behavior of a bridge was defined in the IEEE 802.1 standard.

But bridges never really had enough ports. This brought about the switches that we use now. They were a development of the bridge and had way more ports, they could send data to only the destination device, and most importantly, they were full duplex. It meant that switches could send and receive data at the same time. But this required more than just switches. It also means that each port on the switch means one collision domain, meaning no more collisions!

Ethernet cables had to be developed further to support this. So they went from having 2 pairs of wires, to having 4 pairs of wires. These are the Ethernet cables that we use in networks today. Straight-through and crossover cables still needed to be used for their different use cases.

Straight through cables were used for connecting:

  • Switches to Routers
  • Switches to PC/Server
  • Hubs to PC/Server

Crossover cables were used for connecting:

  • Switches to Switches
  • Switches to Hubs
  • Hubs to Hubs
  • Routers to Routers
  • PC/Server to PC/Server
  • Routers to PC/Server

But technology is all about reducing complexity, not making more of it. So Automatic Medium Dependent Interface – Crossover (Auto-MDI-X) was developed to automatically detect what devices are being used, and what wires should be used depending on those devices. In short, it means it no longer matters if you are using a straight through, or crossover cable. Any Ethernet cable can be plugged into any set of devices, and work perfectly!

Move forward today, and switches are used for far more than just forwarding data based on MAC addresses. They’re used for security, quality of service and even layer 3 routing! And nowadays, most enterprise networks use the 10GBASE-T Ethernet standard, that’s 10 gigabit per second over very similar copper wires we’ve been using for over 40 years!