CIS 106a: Introduction to Operating System Concepts

Chapter 8: Windows on a Network

 

Objectives:

This chapter is about networks and Windows. Objectives important to this chapter are:

  1. Physical network architectures (topologies)
  2. Networking with Windows
  3. Configuring network cards and protocols with Windows
  4. Sharing resources with Windows
  5. Wireless networks
  6. Troubleshooting network connections
Concepts:

This chapter presents an overview of many networking concepts. It is difficult to accept the idea that this material can be adequately presented in so short a span of pages. In keeping with the plan of presenting lecture notes on the chapter, I will discuss the concepts in it here, but I will also note that a longer, fuller discussion of these concepts can be found in my notes for NET 121, NET 222, and CNE 250. We normally spend an entire course on what your text tries to cover in one chapter.

Networking can be defined as users sharing resources (like printers or files) across a common medium (like copper wire or fiber optic cable) by way of specific rules (like TCP/IP or other network protocols). Protocol has two meanings: a rule used on networks, or a program that implements a rule.

In general, networks use hardware and software to accomplish the goals of sharing resources. Information is sent across networks from one device to another. Some networks call devices nodes and some call them hosts. These words apply to workstations as well as to devices like servers and printers.

The author provides a list of acronyms that are useful for judging the relative size of a network:

  • PAN - a personal area network (not a common term) would be a collection of personal devices that are linked together
  • LAN - a local area network is a network linking several devices that may be used by different people, such as several computers in a home, a dorm, an office, or a classroom. A LAN may extend across several buildings.
  • MAN - a metropolitan area network covers a major city. Users on the state of Michigan equipment at the state capital are connected through the LMAN (Lansing Metropolitan Area Network)
  • WAN - a wide area network covers any area larger than one city, such as the Internet which covers most of this planet

Networks exist to provide and share information. Different types of networks have been developed to carry voice, text, video, and other types of data. Networks can be compared by some common measure such as their bandwidth. Bandwidth is a measure of how much data can flow through a network at once. Voice networks (like typical telephone systems) carry analog signals, and their bandwidth can be measured in Hertz (cycles per second). Digital networks, like the Internet, are typically measured in bits per second (or kilobits or megabits per second).

The chapter introduces networking hardware by discussing Network Interface Cards NICs). NICs are typically designed for use on only one kind of network. For example, you can't use an FDDI card on an Ethernet, or a Token Ring card on FDDI. A driver should come with any card you buy. The card and the driver are specific to the network type, but the other software on a workstation is not. MS Office, for example, does not need to be told which network architecture you are using.

Most NICs have physical addresses burned in at the factory. These addresses are called MAC addresses since they work on the Media Access Control sub-layer of the ISO network model. This address provides a unique identifier for every device that has a NIC installed. The device's MAC address may be the only address needed in a LAN, if signals do not need to go to other networks. A MAC address is 48 bits long, which means that it is 6 bytes long. MAC addresses are usually written as hexadecimal numbers, which saves a lot of space: a byte can be written as 8 bits, but it can also be written as 2 hex characters. Six bytes means 12 hex characters in a MAC address. The numbers in the first three bytes identify the NIC's manufacturer. For example, a MAC address might look like this: 00-16-C7-CA-68-7F. Checking the manufacturer at this web site, we find that this NIC was made by Cisco.

When buying a NIC, you should pay attention to three major choice factors:

  • What kind of network does it work on? I would add that you should check the rated bandwidth of the card as well. It in nice to know that it is for an Ethernet, but you also want to know whether it supports 10 Mbps, 100 Mbps, both, or higher data rates.
  • What kind of medium is it for? UTP, coax, fiber optic, etc. The card has to be able to use the kind of cable your network uses.
  • What kind of bus does it plug into? PCI, SCSI, ISA, etc. The card has to fit a slot in your computer, printer, etc.

The text describes several types of cable media:

  • twisted pair - come in two types:
    • unshielded - UTP does not have an EMI resistant sheath
    • shielded - STP has an EMI resistant sheath
  • coaxial - Coax similar to that used for cable TV, but NOT identical; now it can be since cable TV companies are now providers of Internet access
  • fiber optic - glass or plastic channels that conduct light, often red laser light

(For the purists among you, I will note that the speed of light through the copper media is about two thirds the speed of light in a vacuum.)

It may be redundant to remind you that wireless media means that there is no cable of any sort between certain parts of the network. (There are still wires inside lots of components).

Radio is the label used for frequencies from 10 KHz to 1 GHz. Several bands are used. Frequencies that are used for networks can be divided into regulated and unregulated frequencies. Only a few frequencies are unregulated in the United States. It is not possible to guarantee error free transmission in the unregulated frequencies. This is because anyone else can broadcast in those frequencies, causing errors in your transmissions. For this reason, broadcasts are usually limited to low power in unregulated bands, to minimize interference.

More on wireless technologies:

  • 802.11 wireless (Wi-Fi or Wireless Fidelity)
    • Types - 802.11g (most popular), 802.11b, 802.11a
    • Two new standards - 802.11k and 802.11r
    • Ad hoc mode - directly links two wireless devices, such as the way two Palm PDAs communicate
    • Access point (AP) - connects wireless device to LAN, such as when you take a wireless laptop to a coffee shop and connect to their LAN to access the Internet
  • WiMAX (802.16 Wireless/802.16d and 802.16e)
    • Used in public hot spots and as a last mile solution
  • Bluetooth: short range standard; e.g., optical mouse

Wireless LAN technology is discussed. The most popular technology mentioned is WiFi, sometimes called Wireless Fidelity. WiFi follows IEEE 802.11 standards. Common standards are 802.11a, b, and g. More variations are under development. The text explains that 802.11g is a common standard. In fact, many laptops sold in the last few years have the technology installed as a standard feature. Although the text tells us that 802.11b has an operating range of about 100 meters, it is more common to find that you must be within 150 feet of an access point to use it.

Another wireless technology you may encounter is called Bluetooth. It is typically used for point to point communication between two devices, and may have an operational range of about 40 feet.

The text turns to a discussion of various kinds of telephone networks. Several types are listed. The reader should be aware that the nature of a telephone network is that it is meant to interconnect with other networks, making it a part of a larger, multiple type, hybrid network.

  • POTS - Plain Old Telephone Service is the most basic kind of wired telephone network.
  • VoIP - Voice over IP networks use IP (Internet Protocol) to pass voice signals over data networks. The voice signals are chopped into pieces, sealed into packets, and the packets are sent across data networks to devices that receive the packets and turn them back into voice signals. It is not effective unless you have access to a wide, broadband connection. Those words are not redundant, by the way. A wide connection means we can push a lot of bits through it. A broadband connection means we can push more than one signal simultaneously.
  • Cellular WAN - a cell phone network works by setting up a series of radio towers (wireless access points) whose operational ranges overlap slightly. Several cellular network standards are mentioned:
    • GSM (Global System for Mobile Communications)
    • CDMA (Code Division Multiple Access)
    • TDMA (Time Division Multiple Access)
    • GPRS (General Packet Radio Service

It is important to note the difference between a cordless phone and a cell phone. A cordless phone just connects by a radio signal to a base station, which is just a wireless access point to the telephone jack it connects to. This is not the same thing as a cell phone, which can potentially connect to any wireless access point in the cellular grid provided by your cellular carrier.

The information that passes across networks is often broken into pieces, which are sent, reassembled, and processed. The pieces can be called packets. Many kinds of packets exist. Most packets include header and trailer information, such as the name of the device sending, the name of the device meant to receive, the type of data, and other information. The way headers, trailers, and the actual data are arranged defines the sort of frame that your network is using. Most networks use several types of frames, so devices need to be configured for the type of frame they are meant to send and receive.

The chapter discusses Ethernets next. There are many kinds of Ethernets, but as a group, Ethernets are one of the most common kinds of networks used. Some basic terms first:

  • Topology is the study of shapes.
  • Networking topology has two meanings:
    • physical topology means how the network is actually wired together
    • logical topology means how the network acts like it is wired together.
  • Dialog - a dialog type tells us something about how data is passed between two devices on a network
    • simplex - data passes in one direction only, like a television transmission
    • half-duplex - data passes in both directions, but only one way at a time, like a walkie-talkie
    • duplex (also called full duplex) - data can pass in both directions at the same time. Cell phones use this kind of dialog.

The definition for Ethernet does not specify its physical topology, so it is possible to wire it several ways. The text mentions a bus and a star as two possibilities.

A physical bus puts all the nodes in a continuous chain, with cable running from node to node to node, or uses a continuous cable that the nodes attach to. Signals put on the bus by any device travel to all other devices. (See the symbolic illustration on the right).
A physical star runs a cable from each node to a concentrator, a device that can receive signals from one node and send the signals to all the other nodes at once. (The cables can run away from the concentrator in all directions, just like rays of light from a star. See the symbolic illustration on the right.)

An example of a concentrator is a hub. In both scenarios described here, any signal put on the network will reach all other devices on the network. An improved version of the hub, called an intelligent hub or a switch, changes that. A switch receives a signal from one device, but only sends it out to the device it is meant for. A hub cannot do this.

The use of switches on an Ethernet is important because a bus-based Ethernet uses a contention method for media access. This means that all nodes connected to the network compete or contend for access to the network. Only one node can use a bus network at any given time. When two nodes send at the same time, their signals create a collision, which keeps both signals from getting through the bus. The two nodes then each calculate a random number, wait that number of nanoseconds, then try sending again. This method is called CSMA/CD: Carrier Sense Multiple Access with Collision Detection. Nodes listen to the wire and do not send unless they sense no traffic on it. All nodes have access at all times, so it is possible that two nodes may try send at the same time. Collisions are detected when they happen, and the nodes each calculate a time to wait before trying again.

When we use switches, the rules change. A switch can receive a signal, and send it only to the device it should go to next. This allows several simultaneous transmissions between devices connected through a star-wired switch.

Regardless of the physical topology used, Ethernets typically run at 10 Mbps, but 100 Mbps is also common. 1000 Mbps is also possible. It is called gigabit Ethernet, and the bandwidth can be abbreviated as 1 Gbps.

Your book describes several network protocols, their cabling methods, and their typical bandwidths. There are more possibilities. Some of these are no longer very common, but you are responsible for knowing some facts about them. In the chart below, segment length refers to the maximum length of cable that can be used in this each situation. Signal strength tends to fade below useful levels if the segment length limits are not observed. Fading across distance is called attenuation.

Ethernet type Bandwidth Media Segment Length
10Base5 10 Mbps Uses 50 Ohm RG-8 or RG-11 coaxial cable, with DB-15 connector, and vampire taps. Segment length is 500 meters
10Base2 10 Mbps Uses 50 Ohm RG-58 coaxial cable,
with T-connectors. (also called BNC connectors)
Segment length is 185 meters (about 200 yards)
10BaseT and 100BaseT 10 Mbps or 100 Mbps Uses Unshielded Twisted Pair (UTP) or Shielded Twisted Pair (STP) cable, with RJ-45 connectors. Segment length is 100 meters
10BaseF, 10BaseFL, 100BaseFL, 100BaseFX, 1000BaseFX 10 Mbps or 100 Mbps or 1 Gbps Uses fiber optic cable, SC or ST connectors. Segment length is 500 meters to two kilometers

The row for 10BaseT and 100BaseT is highlighted above because it is the variety you are most likely to encounter.

Each of these implementations above uses a three part name: 10 or 100 means the bandwidth in Megabits per second. Base means baseband, which means that only one signal can be on the medium at any time. The third part is meant to be a clue about the length of a segment. A segment is the total length of cable connecting all the devices in a bus configuration, but it is only the length of cable from a node to a hub or switch in a star configuration.

The pictures in your book of cables and connectors are not as good as in other chapters. Look for samples on the Internet for more accurate depictions. Hover over the pictures below for additional balloon notes about them.

The example on the right shows a typical T-connector with BNC fittings. This kind of connector is used with coaxial cable. The fitting on the bottom of the image might attach to a port on a NIC that looks like the barrel on either end of the top of the T. Attachment is achieved by pushing the connector onto the barrel of the port, then twisting the collar of the connector to lock onto the pin that is part of the port. In other words, it mounts like a bayonet.

The next (enlarged) picture shows a BNC connector attached to a thin Ethernet cable. Such a connector would be used to attach to one of the T-connector barrels in the photo above. The other end of the cable would run to the next node on the network.

Your text does not show an RJ-45 connector very well. In the enlarged picture on the right, note the eight gold-colored connections for the eight wires usually found in UTP (Unshielded Twisted Pair) cables. The wires are used in pairs to form circuits. See the NET 121 notes for more information on this sort of connector.

A segment is one part of a network. Most networks can have multiple segments, and still be one logical network.

Regarding segment length, if a signal must travel farther than allowed by the maximum segment length of its medium, you must use a device to strengthen the signal and send that signal to a new segment. An amplifying repeater will do this, but it will also strengthen any static or line noise. A signal regenerating repeater will clean up the signal before sending it out at full strength.

Token Ring networks are like bus networks that connect in a circle. Signals are meant to flow from one device to the next until a signal returns to the device that sent it. The devices take turns instead of contending for access. A special frame is passed from device to device that tells them which device's turn it is to transmit. This frame is called the token.

The token, like all traffic on the ring, passes in one direction only. When a device receives only the token, it becomes that device's turn to transmit. It sends its signal out, attached to the token. Each device on the net receives the signal, one node at a time. If the signal is not meant for it, the node sends the signal on to the next device. When the signal arrives at the intended recipient, that node reverses two bits in the token, to acknowledge receipt of the message. The reversed bits are noticed when the signal is finally returned to the initial sender. The sender then passes the token on.

Token Rings need not be wired as physical rings. A star wired ring is the most common type. Several workstations may be connected to Multistation Access Units (MSAUs), which act like concentrators. The MSAUs are connected together by way of special ports called Ring In and Ring Out. You connect the Ring In port of one MSAU to the Ring Out port of another MSAU. This allows you to extend the circle to include more MSAUs and more workstations as necessary.

The next topic is FDDI, a fiber optic ring standard. This is an ANSI standard, not an IEEE standard, but it makes use of the IEEE 802.2 and 802.5 standards. It is very fast, and has high capacity, making it useful for three main applications:

  • Backbones - connections to other networks that need to be fast and wide
  • Computer room networks - fast connections between critical devices
  • High data rate LANs - connections for users of data intensive applications like CAD

FDDI uses two rings that are counter rotating. This means that traffic travels clockwise on one ring and counterclockwise on the other, making reconfiguration simple. If a break occurs between two workstations, the rings cross over at those workstations, turning the two rings into one, longer loop. (Mouse over the picture to see this happen when a nasty bug breaks the rings.)

Some critical terms appear in the chapter:

  • Server - not just a computer, but the set of hardware and software used to provide a service.
  • Client - any entity on the network that requests a service
  • Peer - a network entity that may request and provide services simultaneously.
  • Workstation - typically, a personal computer that is attached to a network
  • Host - any device assigned an IP address on a network
  • Network Operating System (NOS) - typically, the OS that is run on servers, that makes the network possible (e.g. Windows 2003 Server, Novell NetWare, Linux)
  • Network client software - like a driver for the kind of network you are on, it runs on workstations and allows the workstation to become a part of the network (Some operating systems come with client software for specific NOSs.)

If entities on a network act as peers, then this is Peer-to-Peer Networking. If entities act in strictly defined roles, as either servers or clients, but not as peers, then this is Server-Centric Networking. Most PC networks are this type. A third term, Enterprise Network, is used to describe a network with some characteristics of each of the other two types.

Most new networks follow a client/server model, which is also a distributed computing model. Clients typically perform some or most of the processing on the network, while servers provide services like data storage, instead of providing all the computing power. Client/server networks are typically easier to upgrade, both on the client side and on the server side.

The text lists four protocol suites (each suite can be dozens of actual programs) used on different kinds of networks. Some networks use several suites at once:

  • TCP/IP - Transmission Control Protocol and Internet Protocol are the two protocols that this suite is named for. This suite is used on most modern networks, and must be used to communicate across the Internet. TCP provides reliable delivery of packets on TCP/IP networks. IP is used by routers, devices that find communication paths to other computers. Finding currently valid paths is necessary because any path may be available or unavailable at any given time.
  • IPX/SPX - Internetwork Packet Exchange and Sequenced Packet Exchange are the protocols that give their names to this suite. It was used on all Novell networks before NetWare 5, and is used on legacy networks that contain older Novell servers. SPX works like TCP in older Novell networks (that use IPX/SPX)
  • NetBEUI - NetBEUI (NetBIOS Extended User Interface) only works inside networks, but it provides delivery and error services on them.
  • AppleTalk - AppleTalk Transaction Protocol (ATP) provides reliable delivery of packets on Apple networks

A protocol suite is also called a protocol stack, because the various programs in it give and receive information from each other as though they were arranged in a vertical stack. Protocols must be loaded into memory to be used, and when they are loaded, they are bound to your NIC.

Devices on a network must be given some kind of identifier. This is usually an address, but it may be a name instead. Four identification standards are listed. They are not mutually exclusive:

  • MAC addresses, as noted above, are typically unique to each NIC. These can be adequate identification inside a network, but they do not identify the network the device is on, so they do not carry enough information to send signals across networks.
  • The purpose of an IP address is to identify each unique node and the network that holds it. On an IP network, each device is known as a host, and every host must have an address. The addresses we usually see are actually IP version 4 addresses. They are numeric addresses, stored as four bytes, which is equal to 32 bits. When we write these addresses, we usually place dots between the bytes, but you must understand that the dots do not exist when the addresses are used.

    Since the IP protocol stack was invented with networking in mind, IP addresses contain two parts: one to identify the address of the network a host is on, and the other part to identify the host itself. Every network is assigned an address which could take up one, two, or three bytes of an IP address, depending on the class of the network (A, B, or C). The remaining byte or bytes are typically used for hosts on networks. Each byte in an IP address will hold a number in the range 0 through 255.
  • Character based names - Human beings often have trouble remembering strings of numbers, so Domain Name Service (DNS) allows us to establish a system of naming computers that is more memorable. DNS allows us to type a web address like www.cnn.com. The browser on your machine sends a request to various DNS servers on the system until it receives an authoritative reply that www.cnn.com stands for 207.25.71.29. It then makes a connection to the web server at that IP address. (Note: this is the correct IP address for CNN as I type this. It may not be correct when you read this. This is the virtue of DNS: if CNN finds it necessary to shift web access duties to another server, they only have to update the DNS database, telling it what address the Domain Name actually stands for now.) A system similar to DNS is Windows Internet Naming System (WINS). I guess Microsoft had to try its own version.
  • Port addresses - As discussed above, this is a classic case of giving a word a new definition when it already has several definitions that you already know. This use of the word port means a labeled address in the working memory of a machine (which could be a workstation, but is usually a server). This address is where a particular program is running, which makes it possible for a client to contact a specific program running on another machine without knowing its memory structure.

Regarding the objective of configuring the IP stack on a Windows computer, be aware that you can access the stack from the Network icon in Control Panel (white icon on the right), or from the Network Connections icon, if you have XP (blue icon on the right). You can also drill down to the IP stack through Device Manager or My Network Places. As I have advised before, the actual appearance of windows and tabs will vary from one workstation to another, so do not assume that the configuration screens you encounter will look exactly like those in the text.

The text provides details on installing a NIC, and configuring settings like the assigned IP address, the default gateway (the router you use to send messages out of your network), and the subnet mask. There are several chapters worth of notes about subnet masks on my web site. For the moment, know that a subnet mask tells a computer which part of an IP address stands for the network, and which part stands for a host.

A router is typically used to connect different networks together. A gateway does this as well, but provides translation services between the rules of dissimilar networks.

The text discusses installing a wireless adapter (WiFi card) in a notebook (laptop). This is not necessary in newer laptops, since they generally have WiFi connection capability built in at the factory. For an older laptop, or for supporting a newer technology, the procedure in the book is needed.

Wireless connections should use security protocols to protect the network from unauthorized connections. Some of them are:

  • Wired Equivalent Privacy (WEP) - an early protocol developed for security on Wireless LANs (WLANs). It is no longer considered to be secure.
  • WiFi Protected Access (WPA) - an improved protocol that addresses shortcomings in WEP. WPA provides for encryption of the keys sent across the WLAN (using TKIP) and for user authentication with Extensible Authentication Protocol (EAP).
  • 802.1x - Uses EAP to force WLAN users to authenticate through a dedicated authentication server.