Build a Network at Home

Build a Network, Connect Yourself

A home network can change the way you compute

Introduction
The Two-PC Network
The Multiple-PC Network
Configuring the Computers
Bigger, better Networks; Terminology

Introduction

Up to a couple years ago, the average home was lucky to have one computer. That one computer did all the household computing chores: the office products, the checkbook balancing, and of course, the games. These days, due to the price and availability of PC's, many homes have 2 or more, and their availability is changing the face of computing. To get all of your computers working together, setting up a small, easy to run network should be almost as necessary as putting the computer together in the first place. The network allows communication between the machines, allowing drag n drop file and printer sharing, application serving, and lightning-fast multiplayer gaming.

This may sound like a lot, but the truth is, the whole process is easy to do, and once you set it up, it doesn't need any maintenance. For most of us, having only two or three computers satisfies our needs, and this document discusses a small, private network, targeted at just this sort of environment.

This article strives to be as complete as possible, and at the same time, user-friendly. The concepts explained here are anything but 'quick and easy,' so for most people, who only need to share files across two PC's, the info here may be a bit overwhelming.

You must first decide what sort of things you need your network for-- and if you really even need a network at all. The setup process can take some time, and it can be frustrating. If all you need the network for is copying some files, you may want to first consider other alternatives. If you have only two PC's, you can "build a network" in about 30 seconds, just by connecting a cable. For more than 2 machines, you'll need more parts.

Building the Two-PC network

This network is by no means difficult to build, but things can get tricky. For the simple, two-PC network, you'll need the following:

2 Computers (running windows in this document, although the concept will work with any operating system) with at least one free ISA or PCI slot each.

2 10BaseT, 2 100BaseT, 2 10/100BaseT, or a combination of one of the previous, plus one 10/100BaseT Network cards. 10Base2 Cards will work, but the cabling setup is a bit different, and the technology is essentially obsolete, so it isn't covered in this document.

An RJ-45 Crossover cable. These can be purchased at any computer store, or you can make one yourself by following this article.

We'll get to the configuration section a bit later on, but first, let's talk about our setup here. If you live alone and/or have only two PC's, and never expect to have more, or have a friend over with a laptop, or have children with their own computers, etc., then the Two-PC network will work fine for you. However, considering the rapid expansion of the users' dependence on the PC, it's likely that within a couple of years, you'll have more than 2 computers. If you're looking toward the future, you'll need to work with the Multiple PC Network.

The Multiple PC Network

The Two-PC network provides you with a quick and simple way of connecting two, and only two PC's. It's a network, because you're connecting two computers through a wire, but it's use and functionality as a network is very limited, because it can only accomodate two machines. Nearly all networks are based on network devices, such as hubs, switches, and routers, spread out across great distances (between cities, in many cases). Adding in network hardware greatly increases your computing capabilities, but also complicates things, and for the most part, these devices are beyond the capabilities of the average user, and well beyond his needs.

Our network will be simple. We won't need routers or switches, we won't incorporate fiber optics, and we won't span cities. We will setup a network that's easy to incorporate, is quite expanable, and will allow you to connect every machine in your house, and provide you with the ability to add in more machines as needed. The priciple difference between the Two-PC network and the Multiple-PC network is the precence of network hardware, mainly, a hub. The hub, as its name denotes, is the center of the network. All packets of data pass through it, and if this single piece of hardware should fail, the entire network goes down. Purchashing a high-quality hub is, therefore, important. The better the quality, the less likely problems will be, and the easier it'll be to troubleshoot probems when they do occur. To build the network, you'll need:

x Number of PC's (we're assuming machines running Windows, but the ideas in the article will work on any operating system). Both machines must have at least one free ISA or PCI slot free.

x Number of network cards-- every machine needs one. Make sure that they're all the same speed: all 10BaseT, all 100BaseT, or a combination of the two, mixed with 10/100BaseT cards.

x Number of RJ-45 network cables, assuming you're using a wired network. Wireless networks don't, of course, need cables. The hardware setup of wirless networks isn't discussed here.

A hub that supports x connections. Consider how many machines you have, how many you may wish to connect at any given time, and how many your neighbors may bring over for those late-night games of Quake. :) If you're building your network on 10Base2 (ThinNet, BNC, etc.), you won't need a hub.

Again, the hub will be the center of your network, and its most important part. It should be in a central location: Longer cables don't work as well as short ones (and 10BaseT cables over 300' won't work well, if at all). In other words, if you've got a 3-story house, don't put the hub in the basement-- stick it in a broom closet on the second floor. Hubs can get warm, so while ventilation isn't essential like it would be for the inside of your PC's case, if you've got a monster hub, you might want to keep it in the open, where air can move around it well.

Run cables to your machines carefully:

Make sure they don't pass through doorways or places where people walk
Make sure that they don't run parallel (preferably nowhere near) flourescent lights (or anything else that emits weird radio frequencies)
Keep them away from AC lines and strong magnets. Electricity, when flowing through wire, produces a magnetic field. This field in turn produces electricity in nearby wires, which can travel down the line and toast your NICs.
Keep plenty of slack in the lines. Stetching can snap strands, or untwist them, resulting in a condition known as crosstalk.

Configuring the Computers

As every network card is different, we'll assume from here on that you've installed your NIC properly, and that it'll work correctly when we setup the software side of things.

Let's first get file and printer sharing working. You'll need to do a fair bit of configuring Windows here, so set some time aside. We're only going to be concerned with what's in your Network Properties, which can be reached both through the Network applet in the Control Panel, and by right-clicking on Network Neighborhood and selecting Properties.

We first need to make sure that your network adapter is listed in the Network Properties window. If it isn't, click Add, select Adapter, and choose your adapter from the lists that appear. You'll be prompted to reboot, but don't do it quite yet.

To work with the adapter, we'll need network clients and protocols. The client is the software that interfaces with you, the user, and the protocols work with the operating system and the client to communicate with the network adapter, and ultimately, the network. Unless you have special needs for your network, you'll need only the Client for Microsoft Networks, File and Printer Sharing for Microsoft Networks, and the protocol named NetBEUI. These three objects will allow you to share files and printers with other computers on your network.

To add these Protocols and Clients, follow the same procedure you followed for adding your Adapter under Network Properties, selecting "Client" or "Protocol," rather than "Adapter" in the appropriate menus.

With the Client for Microsoft Networks and the NetBEUI protocol added, you're ready to begin sharing your files and printers. First though, make sure that both of your computers have unique names and are in the same workgroup. In the Network Properties window, click on the Identification tab, and enter the requested information. Make your Computer and Workgroup names one word, with no special characters. In order for both machines to "see" each other through Network Neighborhood, it's usually necessary (and always helpful) for them to be in the same workgroup. After making these changes and rebooting, both computers should be viewable in Network Neighborhood.

Once you've played with these settings a bit, and both computers are viewable, you can start sharing. To share files, they must be located in a folder that you specify as shared. To do this, right click on the folder of your choice, and select "Sharing." (If Sharing is not an option, you must first install File and Printer Sharing: go to the Network Properties box, click "Add," select "Service," "Microsoft," and "File and Printer Sharing for Microsoft Networks.") From the box that pops up, you'll have a few options:

Share Name: The name of your Share (the folder you're sharing). This is what will appear when you browse through the sharing computer in the client computer's Network Neighborhood.

Comment: A fairly unimportant field for home use, as it simply provides a field for you to enter some information about the share for other users on the network.

Access: You have two options, Read only and Full. Read only allows users to view files in the folder, but not write to them. Full allows them to view, modify, and delete files.

Passwords: Here, you can set passwords on the shares. For example, you may want a folder that you can save files to from another point on the network, but you don't want your kids to be able to delete files from the folder. By setting a Read Only password, you can allow certain users permission to view the contents of the share, but not modify them. Users with the Full access password will be able to read/modify/delete the shared items.

You can follow this same process to share printers. Open up your Printers folder, right click on the printer you want to share, and fill in the same info you did to share a folder. Easy, eh?

At this time, I feel it's important to talk about the inescapable topic-- security. We all have to deal with it, and the more we share with the outside world, the more we have to deal with. If you're just setting up a private network at home, you probably don't need to be concerned with malicious users on the network doing anything to your machine. Many of us, however, are on University or corporate networks. Whatever your situation, consider your environment carefully. Set strong passwords: NEVER choose a word that's in the dictionary, no matter what language. Use numbers and characters mixed with letters. Be sure to write down the password in one place, and memorize it for general use (reading the password off of the paper while in the office isn't secure :)). Here's a hint about security: if you want to protect your data, don't make it available. There is no such thing as a password that can't be guessed or cracked. There is no such thing as a system that can't be broken into.

So, we've mastered the simple file and sharing, and you're ready for bigger and better things, eh? Well, continue on...

Bigger, better networks

Our simple file and printer sharing model works fine for your average quick file sharing, but this isn't sufficient for many. Many of us are on large networks with hundreds of other computers, or we want to play IPX games between our computers. Not a problem, provided you've got the time to read a bit.

To get a better understanding of the powers at work here, we need to first discuss the protocols and network engineering at work on most networks. Networks are complex beings, with many underlying principles at work. We speak of protocols and clients, but what to they really do? What's a server? What IS a network? Let's have a look:

Network: Simply, an interconnection of points or nodes (usually computers) connected by communication paths. In our network, the computers are nodes, and the cables we string between them are the communication paths. Of course, networks get incredibly complex, but even as they do, the general idea is the same: nodes and communication pathways. (Note that the Internet is just one huge network.) Networks come in many different ypes or setups, which are usually refered to as topologies.

Protocol: A protocol is a special set of rules that determine how two points on a network communicate with each other. In our simple file sharing example above, we installed the NetBeaui protocol on both (or all) computers on the network; both points must be using the same protocol to communicate properly. Thus, it's essential that protocols are standardized. IPX, TCP/IP, and NetBEUI are examples of network protocols, and no matter what computer you use, the protocols are the same, and will allow you to communicate with another computer on the network using the same protocols. (Note: certain protocols aren't sufficient for communication outside certain parts of the network, but we'll get to this later).

Packets: Data travelling through the network, is sent in small chunks called packets. Each packet contains the internet address (according to the protocol in use) of the node it's destined for.

Client: Clients allow the computer user to connect to the network, using network protocols, and communicate with other users. The client handles the dissasembly of packets for shipment across the network, and re-assembly of incomnig packets into coherent data, such as the GIF you download as part of a webpage.

Network Adapter: The adapter is the physical hardware that interfaces with the network and the computer. All packets are controlled by the adapter.

Network Hardware: This is a broad category, that contains many devices that we aren't concerned with directly, and some that do effect us quite a bit. These devices include:

Hubs: These are simple, easy-to-use devices that allow a number of computers to network with each other, and to be connected to other network segments.
Switches: A Network Switch receives incoming data similarily to a hub, but sends the data directly to its destination: node to node. Hubs on the other hand, take on packet of data, and broadcast to all nodes. Switches, needless to say, are more efficient on network resources.
Routers: Again, networks are made up of smaller networks and network segments. Connecting these smaller networks are devices called routers. Routers simply take packets from one network or segment of a network, and pass it to another segment or network. Routing becomes an important issue when choosing a protocol for your network.
Bridges: A bridge is a simple router. It routes packets between network segments, but can't distinguish bewtween different types of packets. In other words, the two networks must be of the same type. Networks, such as the gigantic internet, are really just smaller interconnected networks.

Client/Server: The client/server model is a simple, yet very important concept in modern networking. The server is generally one machine (but it could be several that work together to appear as one) that provides services across the network to other machines, called clients.

With these terms in mind, we can move on a bit. On a large network, things aren't nearly as simple as our 2-PC, crossover cable network. Most networks have needs far beyond simple file sharing-- access to the Internet. This is done through the TCP/IP protocol. TCP (Transfer Control Protocol) manages splitting data into manageable packets that are transferred over the Internet (or local network) to the destination machine, which uses TCP to reassemble the packets. IP (Internet Protocol) handles the way the data is addressed and labeled.

TCP/IP has limitations though, which can complicate the way we network our computers. Let's take a closer look at IP: An IP address uniquely identifies a computer on the network, making it distinguishable to any other computer in the world (With some specific exceptions, which we'll cover later). It consists of 4 8-bit numbers seperated by periods (.). Example: Avenir Web's IP address is 139.102.48.151. See any limitations yet? The problem lies within the type of number that we're forced to deal with-- 8-bit numbers range from 0 to 255. On most networks, 0, 1, and 255 are reserved numbers, so no ordinary computers on the network can use them. That leaves us with 253 numbers per 'octet.' So theoretically, we could have avenir2.dhs.org at 139.102.48.152, and avenir3.dhs.org at 139.102.48.153, and so on, up to 254, and downward to 2. And after that? Well, once we've filled up our 4th octet, the only thing we can do is start changing the other octets. So maybe we'd have 139.102.49.151 for an IP address. Here's where a minor complexity comes in: When we switch to 48 to 49, we're switching 'subnets' (short for 'subnetwork'). The IP address is broken up as such: domainand.networknumber.subnet.computer. Changing subnets is essentially the same thing as switching network segments, according to the ethernet protocol, which means we need a router to connect these two segments. The router takes packets from the 48 network and passes them to the 49 network, and vice-versa.

Routers work by storing tables of information about what computers and devices are on either side of them. This relative positioning is generally referred to as something being 'upstream' and 'downstream' from the router. When the router (which usually has the 'subnet.1' IP address) receives a request for say, a node with the address w.x.y.z, the router looks through its tables to determine which route the packet should take. If it knows that the data goes to computer z, it will send the packet to subnet y. The same principle works for higher-level routers, where entire large networks meet the internet at a single point.

...

With that little bit of background behind us, we can come to the reason why that's all important. Depending on the size of your network, you may have to choose not only what type of network to implement, byt also what type of network hardware needs to be in place to ensure proper operation. Once the hardware is working, you have to choose what protocols to use. This can be a complex situation. Let's go through the protocols:

NetBEUI: Very simple, fast, easy to use protocol. Also just as limited-- NetBEUI is a non-routable protocol, so packets of this type can't be passed through a router unless 'encapsulated' by packets of another protocol. In other words, NetBEUI can be used to network machines on the same subnet of a network, but for other subnets to see the machine, a different protocol must be used.

IPX/SPX: Somewhat complex, connectionless protocol. Created by Novell, it is used to connect clients to Netware servers and other clients. In a large, non-IP-based network, IPX can be used to share files, etc. between clients.

TCP/IP: Very complex, complete protocol. Most of the Internet runs on it, as it's routable, and highly supported by almost all platforms. Data is transferred from node to node by means of IP addresses, which consist of 4 octets.
Due to the complexity of this document, things are getting a bit confusing. It's all going to eventually come down to a complete re-write, but for now, drop me a line at webmaster@avenir.dhs.org with any comments or suggestions that come to mind. I'll add them in as I get them, and then eventually put it all together in a rewrite of the howto.