Several of our customers have requested a tip explaining the differences between network devices such as hubs, routers and switches. So, here goes:

Bus Cable – The early Ethernet LANs used a coaxial cable that went from one system to the next and connected them all together. Any signal that showed up on the bus could be seen by every computer, and it was about the simplest connection you could get. At least it was simple on paper. It got cumbersome every quickly when the network included more than a few machines or if machines were added or taken out very often. It’s pretty rare these days to find a LAN still using a coax bus.

Hub – The logistics of adding and removing systems is greatly simplified with a hub. At its most basic, a hub is just a box with plugs, called ports, for two or more cables. Inside the box, the signals are transmitted so that what comes in one cable will go out on all of them and be seen by every system. In this regard it is no different than the bus cable. The beauty of it though is that you can plug in a new cable anytime, or take one out, without affecting the connections of the rest.

Going beyond the basics, most hubs will have an indicator light for each cable showing that it has a good connection, and also a light showing when two systems try to put data out at the same time, resulting in a data ‘collision’. And some hubs, called ‘active’ hubs, will amplify the signals going through them. Hubs that don’t amplify are called ‘passive’ hubs. For more complex network arrangements, it‘s possible to use one or more of the hub’s ports to connect to other hubs. In fact, you can buy hubs where the ports are divided into groups, called segments, that are connected internally. And if the administrator can remotely control these connections to isolate particular segments, we have what’s called an ‘intelligent’ hub.

Repeater – This is a specialized type of hub that has just two ports. Its purpose is to amplify the signals between two networks or segments of a network that are too far apart. The ‘passive’ version amplifies everything, including noise. An ‘active’ repeater cleans up the signal before passing it on.

Bridge – Like a repeater, a bridge has just two ports and is used to connect two groups of computers. The difference is that a bridge has a list of which computers are on each side, and it only passes along packets that need to go to the other side. This cuts down on network activity where the two groups have heavy traffic amongst themselves but not too much with each other.

Switch – A switch is like a hub in that it has multiple ports, but internally it has a table listing which systems are found on each port, and shuttles the packets where they need to go. Unlike a hub, signals are transmitted only to the port where they need to go, rather than to all ports. The table (and the network) needs to be simple enough that there is only one possible path for any given packet. As a general rule, switches will be faster than hubs because each port doesn’t have to carry all of the traffic. They also tend to be faster than bridges, and more expensive than either hubs or bridges.

Some switches and bridges can be used to connect networks that use different physical protocols, for instance connecting Ethernet with Token Ring. The name for this is a ‘translating’ bridge or switch. These are specialized devices and not used for general connections in most networks.

Router – A router is an intelligent packet sorter, which can look at the ultimate destination for a packet and analyze the best way to get it there. Where the devices above supply a physical path for the packets, a router also provides the information on how to get from one place to another, and this information is added to the packet header. This makes it a much more powerful device for use with complex networks, including the Internet. In fact, the Internet itself could be described as a network of routers.

A router is often a full-fledged computer system with multiple network cards and its own operating system. The types of tables and routing protocols used by different routers is beyond the scope of this tip, except to say that any router accessing the Internet must use at least TCP/IP.

Brouter – This combines the function of a router and a bridge. Local packets can be routed from one side of the network to the other based on the destination address, even if they do not conform to any routing protocols. Packets with the appropriate protocols can be sent on their way to the outside world.

Gateway – Most of the confusion about gateways comes from the fact that the term describes two different functions. One type of gateway connects a network with another device or network that is radically different, for instance a network of PCs that accesses an IBM mainframe system. A more common use of the term applies to the node on a network that provides access to the Internet, or to other domains in a complex LAN. In networks that have more than one gateway, a workstation will have one of them designated as the ‘default gateway’ to use if no gateway is otherwise specified.

To add to the confusion, in earlier days a gateway referred to what we now call a router, and in fact the gateway and router functions are often combined in the same device. Depending on the situation though, a gateway could be a switch or bridge rather than a router.

Proxy Server – This is a system that sits in between a workstation and the server that workstation is accessing. The most common example is for Internet access, where the browser on a workstation is addressing a server somewhere out on the web, but is actually talking to the local proxy server.

There are a couple of advantages to this arrangement. For one, the proxy server will cache recently accessed web pages, and the workstation can get them faster from the proxy’s cache than it could from a remote site. This speeds up Internet access for everyone. The proxy server can also be used to control Internet access, which is a major issue at some companies. It is a very convenient place to put a firewall too, and it is not unusual to find the gateway, firewall and proxy server functions all combined in the same box, controlling traffic both into and out of the local area network.

Another advantage is that the local network only requires one IP address from the Internet Service Provider, and that’s used by the proxy server while all of the workstations that connect to it use one of the 3 ranges of ‘local’ IP addresses. Given the popularity of the Internet, it’s likely that without proxy servers we would be running short of IP addresses already.

So, that’s the short course on network devices. Thank you to the readers who sent in that suggestion. For the rest of you, what would you like to hear about next?