What is a Hub?

Think of a hub as a rather naïve device- it operates on the Physical Layer of the OSI model. Switches, on the other hand, are much more intelligent- and operate on the Data Link Layer of the OSI model.

When a hub receives information on one port, the information is then broadcasted to all of its ports. This wouldn’t be such a bad plan if it didn’t waste bandwidth and cause collisions. Imagine if two computers transmitted data at the same time: the packets of information would collide and the information would be corrupted. We would have to rebroadcast the data through the process of Ethernet Carrier Sense Multiple Access with Collision Detection- but you can just call it CSMA/CD. In simple terms, it is the protocol we use to resend data after a collision occurs.

Collisions are obviously a problem with hubs, but we just hit the tip of the iceberg. More importantly, hubs are notorious for wasting bandwidth. Hubs operate at half-duplex, which means data can only flow one way at a time. (As compared to full-duplex, where we can send and receive data between two devices at the same time.) Since we are operating at half-duplex, the bandwidth must be shared between each port on the hub. Imagine you have a 20 port hub with a 20Kb/s line to share. Tough luck, but you’ll only get 1Kb/s to each computer on the network.

What is a Switch?

A switch operates on the Data Link Layer of the OSI model. This means that switches are fairly smarter than hubs, as they can route data on a dynamic level. If information is destined for computer A, for example, the switch will only send the data to computer A.

This addresses our collision problem, since switches use what is called microsegmentation. Microsegmentation will allow for multiple collision domains. Which would you rather have- one collision domain or many?

Let’s look at an example. In the below diagram, there are many collision domains (or segments) for the switch network. If computer A and computer B sent data to each other at the same time, for instance, you may have a collision. Computer A and computer C or D, however, will not experience a collision in the process. In the hub network, there is just one collision domain. That means if the first computer wants to transmit data, it could be interrupted by any of the other computers on the network.

But the fun doesn’t stop there! Switches can keep track of which computer address belongs to a certain port. If there is information destined for computer A, it will only travel through computer A’s port. Remember how the hub split the bandwidth between each port? Microsegmentation allows for us to keep the bandwidth as high as possible for each computer. If you have a 20Kb/s line coming in, each computer can have the full 20Kb/s. (But note that switches aren’t magic, if two or more computers are using the line at the same time, they must share it. Still- this is much better than hubs, which automatically split up bandwidth to ports that you might not even be using!)

Should You Upgrade to a Switch?

The short answer here is: definitely. Hubs were primarily used because they are inexpensive and easy to setup. The good thing about “dumb” devices is that they don’t require much configuration or maintenance.

Switches are a bit more expensive, and allow for more configuration options. They can be programmed much like routers, something hubs can’t accomplish. The average home user likely won’t need the extra configuration options- but they will appreciate the speed. If your network is already blazing fast, you may not need the upgrade. If you could appreciate some extra performance, or if you have many computers on the same network, then the upgrade is almost mandatory.

How the Network Interface Card Works

The network interface card operates on layer two of the OSI model- the Data Link Layer. They reside on this layer because they work with media access control addresses (abbreviated as MAC). Every single NIC has a unique number on it. This number is burned into the card at the factory in which it is made, and it can’t be changed. This is to ensure that every computer that connects to a network will be unique- and thus, no “Who’s who” problems will exist.

The MAC address of a NIC is displayed as six two-digit hexadecimal numbers, separated by colons. And although we do need a unique MAC address for each computer, it often doesn’t matter since logical addresses (Such as an IP address) are almost always used, as compared to physical addresses, on large networks.

What to Look for in a NIC

There are three primary things to look for in a network interface card: the bandwidth it supports, the type of media is supports, and the type of network architecture it supports.

Bandwidth

Bandwidth is the term we use to describe how much information can flow through a network connection at a given period of time. You can think of it as a pipe in which water is flowing through. We would of course want the larger pipe, so that we may have more water at a faster rate.

When we refer to bandwidth, we generally group it into four different units: bits, kilobits, megabits, and gigabits. If you are using 10BASE-TX Ethernet cable, you are getting 100 Mbps (theoretically). If your NIC card only supports 10 Mbps, you wouldn’t be able to take advantage of the cable’s speed. Obviously, buying such a NIC would be a poor decision.

Type of Media

Some NICs are only able to work on certain types of media. A NIC that only works on Ethernet cable, for instance, will not be able to use wireless or fiber optic media. It will instead be limited to copper connections.

Some NICs can double over as a two cards in one- it isn’t uncommon to see a NIC these days with an optional wireless component. However, it is still important that the type of media your network is using corresponds to the type of media your NIC card agrees with- or else you’ll be searching for that receipt.

Network Architecture

Lastly, we have network architecture. This is both the physical and logical design of the network. Cables, hardware devices, network topology, protocols, and other equipment all fall into this category. The NIC will, in this instance, need to be compatible with everything else on the network.

This category is kind of a given- but Cisco still defines network architecture as one of the three primary things a NIC card buyer needs to know. And hey- if it’s going to be on the exam, we might as well commit it to memory.

When to Replace a NIC

Replacing a network interface card is usually common sense. Cisco likes to go into specifics, however, and demands that students know exactly at what point the replacement should take place. Generally, you should replace the NIC if you think the NIC is damaged. You may also add a secondary NIC for backup. Lastly, you’ll need to replace a NIC if your network changes and your current NIC doesn’t have the specifications to make the transition (example being a transition from copper media to wireless).