The Physical Network

The Physical Network
In essence, a network is a group of computers or devices connected together by communication links. In networking terms, every computer or device (printers, routers, switches, and so on) connected to the network is called a node. Nodes are connected by links, which could be cables or wireless links (such as infrared or radio signals), and they can interact with any other node by transmitting messages over the network.
We can differentiate networks according to their size:
A LAN, or Local Area Network, connects nodes over a limited area. This area can be as large as the site of a big company, or as small as connected computers in someone's home. The most commonly used LAN technology is the Ethernet network (see next section).
WAN is the acronym for Wide Area Network. Multiple LAN sites are connected together by a WAN. WAN technologies that you might know of include Frame Relays, T1 lines, ISDN (Integrated Services Digital Network), X.25, and ATM (Asynchronous Transfer Monitor). In the next section, we'll further discuss the means of connecting to a WAN.
A MAN, or Metropolitan Area Network, is very similar to a WAN in that it connects multiple LANs. However, a MAN restricts the area of the network to a city or suburb. MANs use high-speed networks to connect the LANs of schools, governments, companies, and so on, by using fast connections to each site, such as fiber optics.
When talking about networks, the term backbone is often used. A backbone is a high-speed network that connects slower networks. A company can use a backbone to connect slower LAN segments. The Internet backbone is built up of high-speed networks that carry WAN traffic. Your Internet provider either connects directly to the Internet backbone, or to a larger provider that connects directly to the Internet backbone.
WAN Lines
To connect to a WAN, there are several options available:
Where a specific customer requires a dedicated network capacity, we can use leased lines. Such lines are usually charged at a flat rate, no matter how much traffic is sent.
Examples of leased lines are Digital Data Service (DDS, running at 2.4 Kbps and 56 Kbps), T1 (1.544 Mbps), and T3 (equivalent to 28 T1 lines).
Switched lines are used by the regular telephone service. A circuit is established between transmitter and receiver for the duration of a call, or data exchange. When the line is no longer needed, it is freed for use by another customer of the network provider.
Examples of switched lines are POTS (Plain Old Telephony Service-standard analog lines that support speeds up to 56 Kbps), ISDN, and DSL (Digital Subscriber Line).
A packet-switching network is where the service provider supplies switching technology to interface with the backbone network. This solution provides increased performance and shares resources between customers, so that bandwidth is available on demand.
Protocols used for switching networks include X.25 (up to 64 Kbps), Frame Relay (up to 44.736 Mbps), and ATM (up to 9.953 Gbps).
Ethernet Protocol
To get a better understanding of how physical networks work, we'll look at the most common LAN protocol, Ethernet. Ninety percent of devices attached to a LAN use the Ethernet protocol, originally developed by Xerox, Digital Equipment, and Intel in 1972. In 1980, the IEEE 802.3, CSMA/CD standard specified a 10 Mbps Ethernet.
Nowadays, Ethernets can support 100Mbps and 1Gbps lines. Many cabling technologies can be employed with an Ethernet. There is a standard naming which indicates the speed of the Ethernet network and the properties of the cable technology in use. Such names start with a number indicating the maximum data transfer speed, followed by a word indicating the transmission technology supported, and finally a number indicating the maximum distance between nodes. For instance, 10Base2 denotes an Ethernet that operates at 10Mbps using baseband transmission, with cables that have a maximum length of 200 meters. Some other common configurations are:
Ethernet Standard
Speed
Typical Cable Type
Description
10Base5
10 Mbps
Coaxial copper
This was the original standard for Ethernet, a so-called thick-net cabling technology.
10BaseT
10 Mbps
Copper
10BaseT is a 10 Mbps network with twisted pair cabling. A twisted pair is simply that-a pair of wires twisted around each other.
100BaseTX
100 Mbps
Copper
100 Mbps with twisted pair cabling and full-duplex (X) capability. Full-duplex means that data can pass in both directions simultaneously.
1000BaseSX
1000 Mbps
Multimode Fiber
1000 Mbps network with fiber optic cables. The S indicates the short wavelength (850 nm) of the laser.
CSMA/CD
Ethernet is a CSMA/CD (Carrier Sense Multiple Access/Collision Detect) network. Multiple devices are connected to the same network, and all have simultaneous access. When a message is sent, it is transported across the complete network as shown in the figure below. The receiver is identified by its unique address, and only this node reads the message; all other nodes ignore it.

There is a potential problem because more than one node could attempt to send a message at the same time, which could result in the packets becoming corrupted. The solution used by Ethernet is that every node monitors the network and is thus aware if traffic is circulating. A node can only start sending data if no data is already being sent over the network. In short, this is the CSMA part of CSMA/CD.
There is still however the possibility that two nodes, after checking that the network is not already in use, start sending a packet at exactly the same time on the same network cable. This would cause a collision between the two packets, resulting in corrupted data. Both senders are aware of the corrupted packet because they still listen to the network while sending data, and thus detect the collision. This is the CD in CSMA/CD. Both nodes then halt their transmissions immediately, and wait a random time before checking the network again to see if it is free to resend the packet.
Every node on the local network uses a MAC (Media Access Control) address for unique identification. This address is defined by the network interface card. A network packet is sent across the network, but if the network card does not identify its host as a receiver, it ignores the packet and passes it on. Incidentally, if the packet is intended for it, it still passes it on, but this time flags it as received. The packet then continues around the network until it gets back to the sender, which can now be sure that the intended recipient received the data.
Other Protocols
IBM developed the Token Ring (IEEE 802.5) network, where nodes are connected in a ring as can be seen in the next picture. With Ethernet, any node can send a message as long as there's no traffic already on the network. With Token Ring, every node has a guaranteed access to the network in a predefined order. A token circulates around the network ring, and only the node that holds the token can send a message. Nowadays, Ethernet is gradually replacing Token Ring networks because Token Ring is more expensive and more difficult to implement.

AppleTalk is a LAN protocol developed by Apple that has been quite popular in schools, factories, and so on.
Asynchronous Transfer Mode (ATM) is another protocol that can be found in LANs. It supports fast network switching and has a guaranteed Quality of Service (QOS), but because the cost of ATM network cards is very high, ATM is a niche player in the LAN market. ATM is only used for LANs in installations that require extremely high performance, for example to transmit medical images such as X-rays between hospitals. In the backbone that drives WAN networks, ATM plays a more important role.
Physical Components
An important aspect of understanding the network is knowing the hardware components. We are going to have a look at the major components of a LAN:
Network Interface Card (NIC)
Hub
Switch
Router
Network Interface Card
The NIC is the adapter card used to connect a device to the LAN. It allows us to send and receive messages to and from the network. A NIC has a unique MAC (media access control) address that provides a unique identification of each device.
The MAC address is a 12-byte hexadecimal number uniquely assigned to an Ethernet network card. This address can be changed by a network driver dynamically (as is the case with DECnet systems, a network developed by Digital Equipment), but usually the MAC address is not changed.
You can find the MAC address of a Windows machine using the command-line utility ipconfig in a DOS box with the /all switch. The next screenshot shows the output produced on my system, where the MAC address is 00-50-DA-E2-2C-97. The first part of this number, 00-50-DA, is assigned to the manufacturer of the network card; the manufacturer uses the remainder to create a unique MAC address:

Hub
Multiple devices can easily be connected with the help of a hub. A hub is a connectivity device that attaches multiple devices to a LAN. Each device typically connects via a UTP (Unshielded Twisted Pair) cable to a port on the hub. You may have already heard about the RJ-45 (Registered Jack-45) connector. This is one of the possible port types on a hub, but a hub can also support other cable types. A hub can have anything from four ports to 24. In a large network, multiple hubs are mounted in a cabinet and support hundreds of connections.

The hub acts as a repeater as it forwards every message from each port to every other port, and to the network. A hub is a fairly simple element of a network, operating at the physical network layer to retransmit data without any processing. This makes hubs easy to install and manage, as they don't require any special configuration.
Switch
Switches separate networks into segments. Compared to a hub, a switch is a more intelligent device. The switch stores the MAC addresses of devices that are connected to its ports in lookup tables. These lookup tables allow the switch to filter network messages, and, unlike the hub, avoid forwarding messages to every port. This eliminates possible collisions, and a better performing network can be achieved. Switching functionality is performed using hardware (through ASIC chips, which stands for Application Specific Integrated Circuit).
As seen in the next picture, switches can be used to connect hubs at a site. If Node A sends a message to Node B, the switch doesn't forward the message to segment 2 because the switch knows that Node B is on the same portion of the network as Node A. However, if Node A sends a message to Node C, the message is forwarded from segment 1 to segment 2.

This sort of arrangement was popular in the early days, when hubs were much cheaper than switches, but it is less common now, as the price of a switch has dropped to pretty much the same as that of a hub. Because of the enhanced network performance from collision reduction, new networks often use switches in place of hubs, and end-users are connected directly to a switch.
Router
A router is an intermediary network device that connects multiple physical networks. With many hosts it can be useful to split a LAN into separate portions, or subnets. The advantages of subnets are:
Performance is improved by reducing broadcasts, which is when a message is sent to all nodes in a network. With subnets, a message is only sent to the nodes in the appropriate subnet.
The capability of restricting users to particular subnets offers security benefits.
Smaller subnets are easier to manage than one large network.
Subnets allow a single network to span several locations.
The next diagram shows how routers might connect several subnets.

If using a router in a LAN, be aware that a router is not as fast as a switch. The router must apply more processing to messages than a switch needs to, and consequently takes a little more time before passing on packets.
Routers are not only used within LANs, they have an important place in WANs where they connect different network lines. The router receives a message and forwards it to the destination using the last known best path to that destination:

A router holds a routing table that lists the ways that particular networks can be reached. There will often be several different routes from one network to another, but one of these will be the best, and it is that one that is described in the routing table. Routers communicate using routing protocols that discover other routers on the network, and support the exchange of information about networks attached to each router.
The information that a router collates about the paths between networks is known as router metrics, and may include information such as packet loss and transmission time. The information used to produce the metrics depends on the routing protocol:
Distance Vector Routing Protocols
RIP (Routing Information Protocol) and IGRP (Interior Gateway Routing Protocol) routing protocols use a hop count, which indicates the number of routers that are passed through on the way to the target network. These protocols prefer paths with fewer routers, regardless of its speed and reliability.
Link State Routing Protocols
The best path calculation of the OSPF (Open Shortest Path First) and BGP (Border Gateway Protocol) routing protocols takes into account multiple factors such as the speed, reliability, and even cost of a path.
Hybrid Routing Protocols
Hybrid routing protocols use a combination of distance vector and link state calculation.
Finding the Route
With TCP/IP configuration, a default gateway can be set up. This is the IP address of the router port that the machine's subnet is connected to. This router is used when a host outside the subnet needs to be contacted.
You can see the local routing table on a Windows system by entering ROUTE PRINT on the command-line. This command displays the gateways that will be used for each network connection. The picture below shows the output for a machine with an IP address of 192.168.0.1 with two network interfaces (one LAN NIC card and one WAN connection). If a host with the address 192.168.0.x is accessed, the local IP address 192.168.0.1 is used as a gateway-we can connect to these hosts directly. For other network destinations, the router 212.183.100.220 is used:

The ROUTE command has an option (ROUTE ADD) to specify the IP address of a router (gateway) and the network address to use with that router. This router will then be used to connect to hosts on the specified network.
Another useful command I want to look at is TRACERT. This allows us to examine the path used to reach a destination. Simply specify the host name or IP address after the TRACERT command- TRACERT www.global knowledge. com in the example below-shows all routers that were used to reach the specified host. As you can see in the picture below, the command also displays the time needed to reach the next hop. This command is very helpful if a host cannot be reached, which could indicate that some network in between is down or not available.