One of the easiest analogies used to understand the OSI system model is that of sending a letter through the mail. A number of items must be completed for your letter to be delivered to the appropriate recipient.We walk a letter through the postal system and illustrate the parallel connections to the OSI system model.

The first thing that you need to do to send a letter is to write it.You sit down at your desk and write a letter to your friend that lives on the other side of the country. After you finish writing the letter, you get an envelope and address it to your friend.You then walk to your mailbox and place the letter inside.These actions correlate to the OSI system model layers nicely.Writing the letter corresponds roughly to the Application layer. If you used a word processor to write the letter, then print it out to place in the envelope, the act of printing the letter would be similar to what happens at the Application layer.The fact that you printed the letter means that you relinquished control of the letter to the network, the postal system in this case.Your actual words on the paper correspond to the Presentation layer in that you needed to ensure that the recipient, your friend, can read the letter.You presented your thoughts in a format your friend can read and comprehend. Addressing the letter can correspond to the Session, Transport, and Network layers. In networking terms, the steps of sealing the letter in the envelope and addressing it relate to the actions of UDP in a TCP/IP network. The data, your letter, was encapsulated in the envelope and passed down through the OSI model to the Network layer where it was addressed.Without the address, your letter cannot be delivered and the same principle applies to networking. Data cannot be delivered without an address. Placing the envelope in the mailbox is comparable to what happens at the Data-link and Physical layers of the OSI system model.The envelope was placed or encapsulated in the correct format for delivery on the network where it will be transmitted to the recipient. The mailbox maps to the Data-link layer and the postal carrier that picks up the envelope would be the Physical layer, responsible for ensuring that the envelope is delivered

Now that the envelope is in the network, the postal system, it may pass through many different offices. If you view these offices as nodes on a network, they would correspond to routers.The envelope reaches your local post office, or default gateway in a TCP/IP network, and is scanned by a computer to determine if the envelope requires routing for proper delivery. In this example, your friend lives across the country, so the envelope does need to be routed.The computers in the post office review the destination address and determine the best path for the envelope to take to reach its final destination.The next office, or hop, on the path the envelope takes may be a regional office or some other central location with routes to the next hop.Your envelope is transported by mail truck, plane, or other form of transportation.The actual path and transmission medium are unimportant to you as you relinquished control of your letter when you placed it in your mailbox.You are trusting that the postal service will ensure that your letter arrives.

Your envelope finally reaches the local post office for your friend.The envelope is delivered to your friend and is opened.Your friend opens the envelope, pulls out the letter, and reads it.These last steps correlate to the OSI system model working in reverse.The data, your letter, is de-encapsulated when the envelope is opened.The contents are then delivered to the recipient when your friend reads the letter, a mapping to the Presentation layer, and comprehends through the Application layer.

The seven layers to the OSI system model are as follows:

• Physical layer
• Data-link layer
• Network layer
• Transport layer
• Session layer
• Presentation layer
• Application layer

We start at the bottom with the Physical layer.The Physical layer of the OSI system model is responsible for defining the electrical and mechanical aspects of networking.Topics such as cabling and the methods for placing the 0’s and 1’s of binary data on the medium are covered in great detail here. Standards such as Category 5, RS-232, and coaxial cable fall within the realm of the Physical layer.

The next layer is the Data-link layer.The Data-link layer defines the protocols that control the Physical layer. Issues such as how the medium is accessed and shared, how devices or stations on the medium are addressed, and how data is framed before transmission on the medium are defined here. Common examples of Data-link layer protocols are Ethernet,Token Ring, FDDI, and PPP.

Within the Data-link layer are two sublayers: the Media Access Control (MAC) and Logical Link Control (LLC).These two sublayers each play an important role in the operation of a network.We start with the MAC first.The MAC sublayer is responsible for uniquely identifying devices on the network. As part of the standards of the OSI system model, when a network interface in a router, switch, PC, server, or other device that connects to a LAN is created, a globally unique 48-bit address is burned into the ROM of the interface.This address must be unique or the network will not operate properly. Each manufacturer of network interfaces has been assigned a range of addresses from the Institute of Electrical and Electronics Engineers (IEEE).The MAC sublayer is considered the lower of the two sublayers and is also responsible for determining the access method to the medium, such as token passing (Token Ring or FDDI) or contention (CSMA/CD). Figure 1.5 shows an example of MAC addresses “on the wire” after being passed from the MAC layer to the Physical layer and being converted to 0’s and 1’s.

The next sublayer is the LLC layer.The LLC sublayer is responsible for handling error control, flow control, framing, and MAC sublayer addressing.The most common LLC protocol is IEEE 802.2, which defines connectionless and connection-oriented variants. IEEE 802.2 defines Service Access Points (SAPs) through a field in the Ethernet,Token Ring, or FDDI frame.Two SAPs are associated with LLC: the Destination Service Access Point (DSAP) and the Source Service Access Point (SSAP).These SAPs in conjunction with the MAC address can uniquely identify the recipient of a frame.Typically LLC is used for protocols such as SNA that do not have a corresponding network layer.

The next layer defined by the OSI reference model is the Network layer.The Network layer is responsible for addressing a network above the Data-link layer. The Network layer is where protocols such as Transmission Control Protocol/Internet Protocol (TCP/IP), Internetwork Packet Exchange (IPX) and AppleTalk tie into the grand scheme of things. Routing functions are also performed at the Network layer.TCP/IP routing protocols such as Routing Information Protocol (RIP), Open Shortest Path First (OSPF), and the Border Gateway Protocol (BGP) operate at the Network layer.We focus more on TCP/IP in the upcoming “Review of TCP/IP Basics” section.

The three previous layers we covered, Physical, Data-link, and Network, are considered the lower level protocols in the OSI reference model.These are the protocols that will more than likely consume the majority of your time as a network engineer. However, that does not mean that the next four layers are not important to the operation of a network.They are equally important, because without the next four layers, your network doesn’t even need to be in existence.

The fourth layer of the OSI system model is the Transport layer.The Transport layer defines the protocols that control the Network layer, similar to the way the Data-link layer controls the Physical layer.The Transport layer specifies a higher level of flow control, error detection, and correction. Protocols such as TCP, User Datagram Protocol (UDP), Sequenced Packet Exchange (SPX), and Name Binding Protocol (NBP) operate at this layer.These protocols may be connection-oriented, such as TCP and SPX, or connectionless, such as UDP.

The fifth layer of the OSI system model is the Session layer.The Session layer is responsible for establishing, managing, and terminating communication sessionsbetween Presentation layer entities and the Transport layer, where needed. Lightweight Directory Access Protocol (LDAP) and Remote Procedure Call (RPC) are examples of Session layer protocols.

The sixth layer of the OSI system model is the Presentation layer.The Presentation layer is responsible for ensuring that data sent from the Application layer of one device is comprehensible by the Application layer of another device. IBM’s Network Basic Input Output System (NetBIOS) and Novell’s NetWare Core Protocol (NCP) are examples of Presentation layer protocols.The ISO also developed a Presentation layer protocol named Abstract Syntax Notation One(ASN.1), which describes data types independent of various computer structures and representation techniques.ASN.1 was at one time thought to be the Presentation layer protocol of choice, when the ISO’s protocol stack was going to sweep the networking industry. Now we know that some components of ISO, such as Intermediate System to Intermediate System (IS-IS) as a routing protocol, and the X.500 directory services protocol have been widely deployed, while the majority of the protocol stack has been neglected.

The seventh, and final, layer of the OSI system model is the Application layer. The Application layer is responsible for providing network services to applications such as e-mail, word processing, and file transfer, which are not implicitly defined in the OSI system model.The Application layer allows developers of software packages to not have to write networking routines into their program. Instead, developers can utilize programming functions to the Application layer and rely upon Layer 7 to provide the networking services they require. Some common examples of Application layer protocols include Simple Mail Transfer Protocol (SMTP), Hypertext Transfer Protocol (HTTP), and Telnet.

In Ethernet, the multiple access (MA) is the terminology for many stations connected to the same cable and having the opportunity to transmit. No device or station on the cable has any priority over any other device or station. All devices or stations on the cable do take turns communicating per the access algorithm to ensure that one device on the LAN does not monopolize the media.

The CS (carrier sense) refers to the process of listening before speaking in an Ethernet network.The carrier sense operation is performed by every device on the network by looking for energy on the media, the electrical carrier. If a carrier exists, the cable is in use, and the device must wait to transmit. Many Ethernet devices maintain a deferral or back-off counter defining the maximum number of attempts the device will make to transmit on the cable. If the deferral counter is exceeded, typically 15 attempts, the frame is discarded.

The CD (collision detect) in Ethernet refers to the capability of the devices on the wire to know when a collision occurs. Collisions in Ethernet happen when two devices transmit data at the same time on the cable. Collisions may be caused by the cable distance being exceeded, a defective device, or a poorly written driver that does not adhere to Ethernet specifications.When a collision is detected, the participants generate a collision enforcement signal.The enforcement signal lasts as long as the smallest Ethernet frame size, 64 bytes.This sizing ensures that all stations know about the collision and do not attempt to transmit during a collision event. After the collision enforcement signal has finished, the medium is again open to communications via the carrier sense protocol.

Deterministic access is the protocol used to control access to the physical medium in a token ring or FDDI network. Deterministic access means that a control system is in place to ensure that each device on the network has an equal opportunity to transmit.

Cabling

The physical infrastructure of a LAN is one of the most important components of a network. If the physical medium that data is traversing is faulty or installed incorrectly, network performance and operation will be impacted. It is analogous to the foundation of a building. Everything in the building is set upon the foundation, typically strong reinforced concrete or other equally durable and reliable building materials. If the foundation is not installed properly, everything built on this foundation is suspect. A LAN is the same, a faulty foundation can be disastrous to a network.You can install all of the high-end gear, switches, routers, servers, but if they don’t have the physical infrastructure to communicate effectively, your network will fail.

There are two primary forms of physical medium a network will utilize: copper and fiber. Between these two forms, there are sometimes many different standards of cable. For example, copper may be shielded, unshielded, twisted, untwisted, solid core, or braided core.We explore copper and fiber cable in more detail to provide a solid understanding of the importance of cabling in your network. You may be asking yourself “Why are we covering cabling in a book on wireless?”That is a very good question.Wireless, as its name implies, does not use physical cabling to provide communications to the wireless network. However, it does use copper cabling to connect to your existing LAN. If your existing LAN has out-of-spec or faulty cabling, your WLAN may not meet your expectations. (Or more importantly, your boss’s expectations!)

The most common form of LAN cabling installed today is copper. Copper has been the “preferred” installation since networks starting taking hold in the corporate world in 1980 when Xerox developed Ethernet. Copper is relatively cheap, easy to install, and can meet most distances that LANs were designed to cover.The original Ethernet specification used what is called thick coaxial cable. This cable lived up to its name for sure! Thick coax is much bigger than the traditional copper cable you might be familiar with. After thick coax came thin coax.Thin coax was a cheaper and easier to handle and install cable alternative. Both of these cable types are implemented in a bus topology.As we covered earlier, a bus topology is linear LAN architecture. Each device or station on a bus is connected to the same medium. One of the major downsides to thick and thin coax was that it created a single point of failure. If the bus were to experience a failure or cut, the network became nonfunctioning.

With the advances made in copper technology, twisted pair cable became a popular LAN medium.There are two main types of twisted pair cable: shielded and unshielded. Shielded, as its name implies, contains smaller copper cables, twisted among themselves with a shielded jacket around them. Shielded twisted pair allows copper cable to be installed in facilities where there is significant interference to the electrical signals passed along the cable.The shielding—as well as the twisting of the cables—plays a role in protecting the able from this interference.Twisted pair cables are less prone to interference than flat, or nontwisted cables.

Among the twisted pair cabling family are a number of different levels of cables.These are commonly referred to as categories, or CAT for short.The primary differences between the categories is the number of twists per foot in the cable. More twists per foot equals less susceptibility to outside interference. Some of the newer, higher categories of cabling also have internal dividers intertwined with the copper cabling to further reduce interference.These higher standards allow faster communications such as Fast Ethernet at 100 Mbps and Gigabit Ethernet at 1000 Mbs over copper cabling.

Bus Topology

A bus topology is a linear LAN architecture in which transmissions from network devices or stations propagate the entire length of the medium and are received by all nodes on the medium.A common example of a bus topology is Ethernet/IEEE 802.3 networks.

Star Topology

A star topology is a LAN architecture in which the devices or stations on a network are connected to a central communications device, such as a hub or switch. Logical bus and ring topologies are often physically implemented in star topologies.

Ring Topology

A ring topology is a LAN architecture in which the devices or stations on a network are connected to each other by unidirectional transmission links to form a single closed loop. Common examples of ring topologies are Token Ring/IEEE 802.5 and FDDI networks.

Mesh Topology

A mesh topology is a LAN architecture is which every device or station on a network is connected to every other device or station. Mesh topologies are expensive to deploy and cumbersome to manage because the number of connections in the network can grow exponentially.The formula used to calculate the number of connections in a fully meshed network is as follows:

(N x (N–1))/2

where N is the number of devices on the network. Divide the result by 2 to avoid double counting the device A-to-device-B connection and the device

B-to-device-A connection.To illustrate the large numbers that a fully meshed environment can reach, review the following examples:

• A small network with 50 users wants to implement a fully meshed topology.The number of connections required to do this would be (50 × (50–1))/2, which equals 1,225.That is a lot of connections for a small LAN!
• A medium network with 500 users wants to implement a fully meshed topology.The number of connections required to do this would be (500 × (500–1))/2 which equals 124,750 connections!

Now for the reality check on fully meshed networks. Fully meshed networks are typically implemented in a small handful of situations.The most common deployment model for fully meshed networks would be in the WAN arena. Frame Relay and ATM are technologies that are well suited for fully meshed networks with high availability requirements. Figure 1.4 depicts a typical mesh network.

There are three primary types of networks, the local area network (LAN), metropolitan area network (MAN), and the wide area network (WAN).The distinguishing feature of these networks is the spatial distance covered. LANs, as the name implies, are typically contained in a single structure or small geographic region. Groups of LANs interconnected may also be referred to as a campus in larger environments. MANs connect points or nodes in a geographic region larger than a LAN, but smaller than a WAN. Some of the same LAN technologies may be employed in a MAN, such as Gigabit Ethernet.WANs are geographically diverse networks and typically use technologies different from LANs or MANs. WANs typically are comprised of high-speed circuits leased from a telecommunications provider to facilitate connectivity.WANs rarely use the same technologies as LANs or MANs.Technologies such as Frame Relay, Integrated Services Digital Network (ISDN), X.25, Asynchronous Transfer Mode (ATM), Digital Subscriber Line (DSL) and others my be used.This is because of the larger distances WANs service.

Where can you fit WLANs into your existing infrastructure? Just about anywhere you like.WLANs allow network designers to no longer be constrained by the 100m distance limitation for Category 5 copper cabling. Because WLANs use radio frequency (RF) signals to communicate, users can stay connected to the network almost anywhere.

Many companies are merging WLANs into their traditional wired networks to provide connectivity to the network to large numbers of users. Conference rooms are a great place to start considering wireless in your network.The cost ofwiring a conference room and maintaining the hardware required to keep those wired jacks “hot” can be prohibitive. Conference rooms are used for “chalk talk” design sessions, application development sessions, and training. By using WLANs, the need for multiple data jacks in a conference room can be eliminated. A single antenna connected to a WLAN access point (AP) can support many users.

Warehouse applications are also prime candidates for WLAN. Real-time inventory control can be implemented using wireless. Imagine having your inventory control software connected to mobile devices on the warehouse floor tracking inventory as it fluctuates during the course of a day.WLANs can be a very important business driver, enabling a company to gain a competitive advantage.

Hospital bedside access is also a popular application for WLANs.The ability for a hospital staff member to check in a patient at bedside rather than waiting in line at an admissions desk is much more efficient. Bedside access can also enable a doctor to write a prescription or check medical records on a patient instantaneously.

College campuses and some companies are also extending the network infrastructure to public access areas both indoors and outside.This no longer restrains the user to just her desk, or even in the building, to be productive. For the growing mobile workforce, wireless provides the connectivity.