Chapter 2. Routing and Bridging Overview
Routing and bridging are the main things that Internetwork Operating System (IOS) does on a Cisco router. You should understand routing and bridging before you begin IOS configuration; therefore, we're going to cover some computer network basics, and then we'll use those basics to do an overview of routing and bridging.
A computer network is a broad term meaning a collection of interconnected, autonomous computers; interconnected computers can exchange information with each other (Tanenbaum 2). I'll be referring to a computer network as simply a network. Networks are separated by special computers called routers or bridges. For the purposes of this book, an internetwork is a collection of networks separated by routers and/or bridges.
Routing and bridging are two different communication mechanisms used to exchange information between computers on different networks, across an internetwork. The mechanism you decide to use will depend on the type of network you have and the protocol that your computers are using to exchange information. Examples of information exchanged on a network and an internetwork are documents, databases, E-mail messages, and World-Wide-Web (WWW or web) pages.
Since most routers can perform both routing and bridging, discussion can be pretty confusing so we need some terminology guidelines. When I refer to a router in this book, I will be talking about a device that is performing routing unless I state otherwise. When I refer to a bridge, I will be talking about either a bridge or a router performing bridging.
When your computer sends information on a network, the computer splits the information into little messages called datagrams. The datagrams get addressed so that, once they get on the network, they arrive at the computer to which your computer is talking. If the other computer is on another network, a router must take the message and get it to the correct network. The router does this by examining the address in the message to determine where the message is supposed to go.
You can think of router operation as being similar to the postal service. When you send a letter to your Aunt Emma in Kansas, you put an address on the outside of the letter's envelope. The postal service examines the address to determine which direction your letter is supposed to go next; in other words, the postal service routes your letter based on the destination address. Of course, the destination address must be correct; otherwise, Aunt Emma will never get the letter. Pretty simple, huh?
To explain routing and bridging, we need to cover some basic concepts of computer networking. The basic concepts of networking that we'll cover include the Open Systems Interconnection (OSI) Reference Model, encapsulation, and computer address format. We'll then take these concepts and put them together to describe how routing and bridging work.
The OSI Reference Model comes from the International Organization of Standardization (ISO) and provides us a standard set of terms that we can use to discuss computer networking concepts. The OSI Reference Model describes the functions of a network as layers, seven of them. The layers are stacked and numbered from 1 to 7, bottom to top, like the stories of a building. The seven-layered model is shown in Figure 2-1.
<<<J101 - Figure 2-1 OSI Reference Model>>>
Starting at the top layer, we will briefly review each layer's function. We are, however, going to concentrate on layers 2 and 3 to describe routing and bridging.
Layer 7, the applications layer, is where all network applications run. These are the applications that you use to send information, data, across the network, for example, web browsers and servers and E-mail clients and servers.
Layer 6, the presentation layer, controls the formatting of data for your application's use. For example, if the data is sound from a radio broadcast to which you are listening across the Internet, the presentation layer formats the data for the application that you're using to receive the broadcast.
Layer 5, the session layer, handles the establishment and termination of communication sessions. For example, when you login to your company's mainframe, a session must be established for you to be able to talk to the mainframe, and, when you logout, your session is terminated.
Layer 4, the transport layer, provides source-to-destination connection establishment for data transfer. Over the connection, the transport layer can provide flow control and error checking.
Layer 3, the network layer, is the most important layer in routing. The network layer contains the address that routing software examines to determine where to route a message across a network.
Layer 2, the data link layer, is dependent on the medium to which a host is attached. The data link layer links the data from the network protocol to the physical medium on which it is to be transmitted.
Layer 1, the physical layer, defines the characteristics that belong to the type of physical network connection on a host, for example, the cable type, the connector type, the signal frequency, the signal level, and the maximum cable length. The physical layer determines exactly how bits (binary ones and zeroes) are transmitted and received on a network.
A computing device that is connected to a network is called a host. For example, your computer and a Cisco router are both hosts. To talk to each other, hosts must abide by a set of rules. The rules of communication used between hosts are called protocols, and many have been developed. Hosts should use the same protocol to share information. Five protocols are covered in this book.
The data link layer, layer 2, is independent of the protocol being run on a host; it changes only with the network medium, for example, Ethernet and token ring. What happens at the upper layers changes from one protocol to the next.
The software implementation of a protocol on a host is called a stack, because it's just a "stack" of functional layers. Each individual protocol may have different names and numbers for its layers, but the functionality of the seven layers exists somewhere within the protocol; therefore, we can use the OSI Reference Model to talk about protocols in general.
Encapsulation is the process of adding headers to data at each layer of a stack. Data that is to be sent over a network starts at the application layer and moves down a protocol stack until it leaves a host at the physical layer. At each layer of a host's stack, a header is placed in front of the data. Your data combined with a header or headers is a datagram. A datagram at layer 3 is called a packet; a packet starts with a layer-3 header. A datagram at layer 2 is a frame; a frame starts with a layer-2 header (frame header).
A datagram header provides a path up a stack toward an application. A header must contain a data field that indicates the type of data encapsulated at the layer immediately above the layer that added the header. For example, when your computer receives web page from a web server, the header that immediately precedes the web-page data must contain a field that indicates to your computer that the data being received is meant for your web browser. Another example occurs when a host is encapsulating a packet with a frame header at layer 2, the frame header must contain a value that indicates what protocol is being spoken; this value could indicate what type of layer-3 header immediately follows the layer-2 header in the frame.
Let's revisit the postal service analogy. Suppose you are mailing a gift to your Uncle Charlie in Seattle. You put the gift in a box. You then wrap shipping paper around the box. Finally, you write Uncle Charlie's address on the shipping paper so the postal service will know how to route the package. You have just encapsulated the gift twice and put an address on the outer wrapping.
Your process of preparing the gift for shipment is analogous to the process a host goes through when it sends data. The data starts off in an application and must move down the stack. As the host is preparing the data for transmission, it encapsulates the data with a header at each layer as the data moves down the stack. The last header contains an address that allows the data to reach its destination.
The header at layer 2 always contains an address; since this is a data link, or frame, header, the address type changes based on the medium on which the frame is to be transmitted. For some protocols, like IP and AppleTalk, the layer-3 header also has an address. If an internetworking host, like a router, examines a frame's layer-2 address to determine where to send a network message, the internetworking device is bridging frames. If an internetworking device uses a layer-3 address to determine where to send a network message, the internetworking device is routing packets.
When Uncle Charlie receives the package you sent, he checks the destination address on the outside to make sure it's his. He removes the shipping paper and, then, opens the box. He now finds out what the gift is.
A host, that receives a network message, reads the destination address to determine if the address is its own. If the address matches, the host moves the data up the stack, de-encapsulating at each layer. Each layer removes a header that was added by the corresponding layer on the transmitting host until all that is left is the original data that was transmitted.
The format of a layer-2 address changes by the class of network that is using the address. Local Area Networks (LAN's) and Wide Area Networks (WAN's) are the two main classes of networks. A LAN is a network with hosts that are directly-connected and close to each other. Control of a LAN usually belongs to a single company. Types of LAN's include Ethernet and token ring. LAN's use a Medium Access Control (MAC) address at layer 2. A WAN is a network with hosts a large distance apart. The installation and administration of a WAN usually requires the assistance and facilities of a telecommunications carrier (for example, a phone company). Some examples of WAN's are frame relay, leased line, and dial-up. A WAN's layer-2 address depends on the type of WAN.
Layer-2 addresses are important because a host must put a destination address (and sometimes a source address) in the frame header during the encapsulation process before a frame can be transmitted.
Most LAN's use MAC addresses at layer 2. The MAC address identifies a host on a LAN, and it allows a frame to properly navigate on a LAN. That is, it is used to get a frame from one host to another on the same LAN. Since LAN's are usually separated by routers and routers usually base their routing decisions on layer-3 addresses, a MAC address has no bearing on how data is routed. If, however, a router has multiple LAN interfaces and is running bridging, the LAN's are treated as one logical, or virtual, LAN. Since bridging causes a router to examine the layer-2 address for message forwarding, MAC addresses become significant in the router's decision to forward a message from one physical LAN to the other.
The format of a MAC address is show in Figure 2-2. As you can see, a MAC address is 48 bits long; that is 48 binary ones and zeroes. These bits are usually written in hexadecimal (hex). Since a hex digit represents four bits, there are 12 hex digits in a MAC address.
The first half of the MAC address is the Organizational Unique Identifier (OUI) which is sometimes referred to as the vendor code. For example, the MAC address of my office computer is 00-A0-24-37-8D-9E. The first six hex digits are 00-A0-24; this is the OUI for 3Com. So you can tell I have a 3Com LAN card in my computer.
The second half of the MAC address is the serial number; a manufacturer assigns a unique serial number to each LAN interface it produces. The combination of OUI and serial number guarantees that MAC addresses are unique to each LAN interface as long as manufacturers follow the guidelines.
Each type of LAN has its own encapsulations, or frame formats. For Ethernet, there are four encapsulations; for token ring, there are three; and for Fiber Distributed Data Interface (FDDI), there are two.
When a LAN host encapsulates a packet with a frame header (in front of the packet) and trailer (at the end of the packet) to create a frame, the host must put both the source MAC address and the destination MAC address into the header. The source MAC address is the host's own MAC address. The destination MAC address is the MAC address of a host on the same physical or logical LAN to which the host is sending the frame.
Since a MAC address is required to send a frame across a LAN, how does a host get the MAC address of another host? There are three ways:
Using ARP, a host broadcasts a request onto a LAN asking for the MAC address of a destination host. If the desired destination host receives the request, it will reply to the source host with a MAC address that can be used to complete the building of a frame. ARP is used in IP and AppleTalk.
With prediction, a host runs another host's layer-3 address through a calculation to figure out what the destination MAC address is. Predictable MAC addresses are used in IPX and DECnet.
Some network protocols require that hosts periodically send Hello messages that contain their layer-3 address and their layer-2 address. Other hosts listen to these Hello messages and store them in a table for future reference. Hellos are used to acquire MAC addresses in Banyan VINES.
As we cover the configuration of the individual protocols, we'll go into more detail about how the MAC address acquisition occurs.
There are three different types of MAC addresses:
A unicast address is one that is unique to a single host. When one host wants to send a message to a specific host, a unicast address is used. My office computer's MAC address I mentioned above is a unicast address.
A broadcast address is used when a host wants to send a message to every host on a LAN. The broadcast MAC address is represented by 48 binary ones or 12 hex F's (0xFFFFFFFFFFFF). A message sent to the broadcast address is processed by every host that receives it.
A multicast address is one that is meant for multiple hosts, but not all hosts, on a LAN. Multicast addresses are used in lieu of sending multiple unicast messages that would take unnecessary LAN resources and broadcast messages that would take unnecessary host resources.
WAN addressing involves just one address in the frame header. The reason for this is that most WAN data paths are point to point. In other words, each data path of a WAN has only two hosts. When a host sends a frame directly across a WAN, there's only one place it can go - the host on the other end of the data path. Likewise, according to the host receiving a frame directly from a WAN, there is only one place the frame could have come from - the host on the other end of the path.
WAN data paths can be either physical or logical (virtual). For example, a dedicated T1 leased line is a physical data path, and a frame-relay Permanent Virtual Circuit (PVC) is a logical data path. When one host wants to talk to another, it doesn't have to identify itself in the frame header. Each host already knows who's on the other side of the link.
Addresses in physical point-to-point WAN's (for example, leased line and dial-up), are host addresses. Addresses in logical point-to-point WAN's like frame relay are data-path addresses.
A layer-3 address appears in the layer-3 header built during the encapsulation process. The layer-3 address is the one that a router uses to perform path determination for those protocols being routed. Not all protocols have an address at layer 3; those protocols that don't have layer-3 addresses must be bridged.
An address at layer 3 consists of two parts: network and node. The network part is the LAN or WAN address. The node part is the address of the host that is attached to the LAN or WAN indicated in the network part of the address.
Consider a building's address consisting of a number and a street name, for example, 1600 Downing Street. The street where the building stands has a name; the street name must be unique in the city. The building has a number; the number doesn't have to be unique in the entire city; it has to be unique only on the street. Think of the street as the network (LAN or WAN), and the building as a node (device) on the network. A network address must be unique in an internetwork, but the node part of the address needs to be unique only on the network. The full host address is the combination of the network part and the node part of the layer-3 address. A host address must be unique in an internetwork. Figure 2-3 shows example layer-3 addresses for three protocols.
The network and node parts of a layer-3 address are normally separated by a period (".") when they are written. The period is verbally communicated with the word "dot." For example, the IP address in Figure 2-3 would be pronounced, "172 dot 16 dot 126 dot 99."
Each protocol that has layer-3 addressing can be routed; we will see and configure these protocols in the coming chapters. A router bases its path determination on the network portion of the destination address in the layer-3 header. Each protocol capable of being routed (a routed protocol) has a different format for its layer-3 address. We will cover each protocol's address format when we cover the protocol.
Now that we have gotten some of the little, but very important, stuff out of the way, let's put it all together for the big picture. What do encapsulation and addressing have to do with routing and bridging? Practically everything.
Figure 2-4 shows the Router Operation Flowchart that will be used in describing some of what a router does when it's either routing or bridging. We will be going through the steps in a little more detail, but keep in mind this is still very basic and does not cover every possibility.
Before a router can attempt to process a message, the router must receive the message, as a frame, on one of its interfaces (Figure 2-4, Box 1). If the interface that received the frame is a LAN interface, the router examines the frame's destination MAC address to determine if the frame is destined for the router (Figure 2-4, Box 2). If the frame is destined for the router, the router will attempt to route it if the encapsulated data's protocol has been configured on the router (Figure 2-4, Box 3). If the frame is not destined for the router, the router will attempt to bridge it if bridging has been configured on the router (Figure 2-4, Box 4) and if the encapsulated data's protocol has not been configured on the router (Figure 2-4, Box 6).
When a host is running a routed protocol and wants to get a message to another host on another network, the host normally sends the frame to the router to be routed. When a host is running a bridged protocol, the host is not aware that a bridge exists; therefore, the host will never send a frame directly to a bridge.
Bridging is the process of forwarding a network message based on an address in the message's layer-2 header. There are many types of bridging supported by IOS:
To illustrate the bridging process, we'll briefly cover transparent bridging. This type of bridging is called transparent because a router running it is transparent to the hosts that are attached to its interface LAN's. Notice that in the Figure 2-4, Box 4, the router checks for bridging only if it receives a frame not destined for it. If the router receives a frame that is not destined for it and the router is not running bridging, the router just discards the frame (Figure 2-4, Box 7). A router (bridge) running transparent bridging performs the following functions:
A bridge processes all frames on all interfaces and dynamically learns the location of each host. When a bridge receives a frame on an interface, it puts an entry into a bridging table that shows the MAC address of the transmitting host and the interface that received the frame. Using this technique, a bridge populates its bridging table for use in path determination. As the bridge is learning the location of each host, it is performing the three F's: Flooding, Forwarding, and Filtering.
If the bridge receives a frame with a destination MAC address not in its bridging table, the bridge will flood the frame out every interface except, of course, the one on which it was received (Figure 2-4, Box 11). This will, hopefully, allow the frame to reach its destination regardless of the destination host's location. When the destination host responds back to the originating host, the bridge will update the bridging table with its location.
Figure 2-5 shows a message sent from Host 1 to Host 2. The encapsulation levels and headers are extremely over-simplified so that we can concentrate our coverage on the addressing of network messages and how that affects bridging. MAC addresses are shown as hex digits A through D.
When Host 1 sends a frame to Host 2, the bridge will find Host 2's MAC address in the bridging table, and the bridge will forward the frame out the appropriate interface if the destination interface is different than the source interface (Figure 2-4, Box 15). Figure 2-5 shows different reception and transmission interfaces.
When frames are forwarded or filtered by a bridge, the are sent exactly as they were received. In Figure 2-5, the frame is forwarded intact out the interface where Host 2 is connected. Note that the bridge is not specifically addressed; it is transparent. Host 1 doesn't know that the bridge is there. Note also that the network address of both LAN's is the same. The two physical LAN's are treated as a single, logical LAN.
Suppose that two hosts on the same LAN are having a conversation and the bridge has learned that they are both out the same interface. The bridge will filter the frames that it receives from the hosts' conversation (Figure 2-4, Box 14). In other words, the bridge will not transmit the frames out other interfaces. This keeps the conversation traffic local and prevents the local conversation from interfering with conversations on other LAN segments.
Bridging does have some drawbacks. A bridge will always flood a frame the first time a MAC address is referenced. That's not too bad; however, a bridge will also flood all broadcast frames and all multicast frames. If your application relies heavily on broadcast or multicast addressing, this could cause considerable network traffic overhead since all broadcast and multicast frames will appear on all LAN segments.
Routing is the process of forwarding a network message based on an address in the message's layer-3 header. There's a lot more involved than just reading a layer-3 address, though. Let's walk through the basic steps that a router performs to forward a network message. The Router Operation Flowchart in Figure 2-4 will be used again.
Figure 2-6 shows the flow of a message from Host 1 to Host 2. The encapsulation levels and headers are extremely over-simplified so that we can concentrate our coverage on the addressing of network messages and how that affects routing. Layer-2 addresses are shown as hex digits A through D.
Remember that a MAC address is used for navigation of a LAN. If a host wants to get a message to another host on the same LAN, the originating host puts the MAC address of the other host in the frame header; otherwise the originating host send the message to a router for routing to the destination network. When a router receives a frame directly addressed to its own interface, the router knows that it should attempt to route the message within the frame. Notice, in Figure 2-6, that the destination MAC address in the frame transmitted by Host 1 is B, the Router's. Encapsulated inside the frame is a layer-3 header that includes the destination network and host addresses.
If the message is to be processed by the router, the router must figure out what protocol the hosts are speaking so it knows how to handle the messages. Remember in the Encapsulation section, I said that each layer's header must contain some value that indicates what is encapsulated at the next layer. The router must check this value in the frame header to find out what protocol is encapsulated in the frame. In the frame header, this could be either a Protocol Type or a Service Access Point (SAP).
If the Protocol Type or SAP field indicates the frame contains a data for a protocol that is being routed by the router (Figure 2-4, Box 3), the router removes the frame header and begins processing what's left, the packet that begins with a layer-3 header (Figure 2-4, Box 5). There is a separate program running on the router for each protocol that is being routed; I'll refer to these programs as routing processes.
The routing process examines the destination network address in the packet header (this is layer 3) and looks for a match in a special table, the routing table (Figure 2-4, Box 9). The router maintains a routing table for each protocol that is being routed. The routing table contains a list of all the router's known networks and their direction.
If the router doesn't find an entry in its routing table that matches the destination network address, the router discards the packet (Figure 2-4, Box 8).
If the router finds a routing table entry that matches the destination network address, the router uses the information in the routing table to determine which interface out which the packet must be forwarded.
Since messages must appear on the network as frames, the router must now create a frame by encapsulating the packet. In other words, the router must put a new frame header and trailer on the packet (Figure 2-4, Box 12). Note in Figure 2-6 that the packet itself does not change.
Since there are many different frame header formats, the router uses the interface name and encapsulation found in the routing table entry to determine what the packet's new frame header is supposed to look like.
In Figure 2-6, the packet is supposed to be forwarded out a LAN interface; therefore, the router must place the destination and source MAC addresses in the frame header. In the new frame that is to cross the LAN from the router to Host 2, the destination MAC address is D, Host 2's, and the source MAC address is C, the Router's transmitting interface.
You may be wondering how the router managed to learn about other networks that are used to populate its routing table. The routing table didn't just appear out of thin air. There are a three ways that a router can have its routing table populated with destination networks. These are as follows:
Networks to which a router is directly connected automatically appear in the routing table.
Networks that a router learns from us are normally called static routes. These are just manual entries into the routing table. We tell the router which direction we want it to route packets destined for a network.
With the proper configuration, routers will talk to each other and share information about networks. Routers talking to each other. Scary. However, configuring your routers to learn about the network from the other routers takes the building of the routing table out of your hands and puts it into the router's. Your job then becomes the monitoring of the routing tables to make sure that the routers are playing nicely with each other. The protocols that routers use to share network information are called routing protocols.
The protocols that carry real data and are "routed" by a router are called routed protocols. Each routed protocol (like IP) that you can configure on a router has at least one associated routing protocol (like RIP and OSPF). The configuration of routing protocols is what I call the fun part of configuring IOS. We will see the routing protocols for each of the routed protocols in the configuration chapters.
You can run as many routed protocols and as many routing protocols as you need or want as long as your router has adequate power and memory. The IOS will create a routing table in memory for each routed protocol that is running.
There are two main types of routing protocols:
The one that's implemented in your network depends on many things such as network requirements, business standards, and even personal preferences.
Distance Vector Routing Protocols
Routers running a distance vector routing protocol learn about other networks directly from their neighboring routers. Distance vector routing is sometimes called "Routing by Rumor" because each router learns about networks from its neighbors' perspective. The routers learn how far away and which direction each network is, that is, each network's distance and vector.
Distance vector routing protocols are pretty easy to spot. The following are some of their characteristics:
A routing running a distance vector routing protocol broadcasts frequent updates out all of its interfaces for its neighboring routers to process, even when nothing has changed in the network. The updates contain all of the networks that the router has in its routing table, unless split horizon is running.
Split horizon is very simple. A router is not to broadcast a path to a network out the same interface in which it learned about the network. In other words, don't repeat to your neighbors what your neighbors just told you.
The updates are transmitted every few seconds. For example, the update period could be every 10 seconds, every 90 seconds. The exact period depends on the individual routing protocol. Routers use an algorithm, sometimes called the Bellman-Ford algorithm, on all of the information it receives in updates. The Bellman-Ford algorithm produces each network's best path, which is placed in the routing table.
Remember the broadcast address? Frames sent to the broadcast address are processed by all hosts; therefore, all hosts on a LAN, with routers that are running a distance vector routing protocol, will be receiving and processing the updates. The hosts will discard the updates only after they have been interrupted from their normal tasks.
Distance vector routing protocols tend to be preferred in small to medium-sized networks. The frequent, periodic updates tend to take too much bandwidth away from the all-important user traffic as a network grows.
Routers running a link state routing protocol learn about networks from the other routers in an internetwork, not just those directly connected to its own networks. Each router transmits a link-state advertisement packet (LSA) to a multicast address. The LSA contains information about the individual router and the links, or networks, to which it is connected.
All of the routers receiving the LSA, process it and then flood it to other routers. This way, all routers running the link state routing protocol receive all the LSA's and, thus, learn about all the networks. All the LSA's received by each router are placed into a database. The router runs a special algorithm against the database to create a map of the network. The algorithm is called the Shortest Path First (SPF) algorithm. The map shows all of the routers and all of the LAN's and WAN's. Based on the map, each router can calculate the best path to each network and can update its routing table.
Updates for link state routing protocols are not transmitted at short, periodic intervals. They are instead triggered by state changes in a router's network connections. For example, if a router's LAN interface goes down, the router would transmit a new LSA indicating that the router no longer has a link to the LAN. Each of the routers receiving this LSA would run the SPF algorithm to create a new network map and then update its routing table.
Since routers base their routing decisions on the network portion of the destination layer-3 address, a routing table contains a list of network addresses.
Layer-3 addressing is hierarchical. By hierarchical, I mean that the address has a general part and a specific part. The network part is general and the node part is specific. The network part is said to be significant for routing since the node part doesn't play a role in route determination.
Our home addresses usually consist of a street address, a city, a state, and a zip code. Suppose that you lived in Dallas and your Uncle Jimmy Mack has mailed you a package from Lexington, Kentucky. When the postal service wants to route the package to you, imagine that they have to look up your exact street name in a very large database to determine where the package is supposed to go. Finding your street in the database could take a long time. All the postal service Lexington really needs to know is the path to Dallas. Once the package gets to the Dallas office of the postal service, the package can be routed to your street.
Consider that Dallas is a big network (city) consisting of many LAN's and WAN's (streets). Each LAN and WAN has some hosts (homes). The routers within Dallas need to know paths to each LAN and WAN, but routers outside Dallas need to know only a general path to the big network.
Maintaining entries for just networks (the general stuff) reduces power and memory load on a router. Routing tables do not normally contain entries for individual hosts.
IOS is the software that you configure on your Cisco router hardware to either route or bridge your information from one network to another. IOS provides the strength of Cisco's router product-line; it's what makes a Cisco router a Cisco router. When you purchase a Cisco router, you must also purchase a license to run IOS. IOS comes in many flavors based on version and feature set. You must make the decision about which flavor you want to run.
There are many versions of IOS. Generally, you decide which version to run based on your comfort with running new software or older, more-tested software, your need to implement a specific IOS feature, or your desire to use a specific Cisco hardware platform.
Cisco uses a special numbering scheme to keep track of IOS versions. The full version number of your IOS has three numbers: major version, minor version, and maintenance release. The major version and minor version numbers are separated by a period ('.') and are referred to, collectively, as the major release. The maintenance release number is shown in parentheses. For example, the IOS version number 11.2(10) refers to maintenance release 10 of major release 11.2. Cisco releases IOS updates often; when they issue an update for IOS, they generally increment a maintenance release number that's associated with the major release number.
Since there are so many versions of IOS, Cisco issues release notes that contain descriptions of release changes and additions. You should read the release notes if you want to find out what has changed in a release or what has been added to a release.
Cisco uses special release designations to let you know how stable they feel the software is. These release designations are as follows: General Deployment (GD), Limited Deployment (LD), and Early Deployment (ED). As a general rule, GD releases of IOS are the most stable. Cisco puts the GD designation on an IOS release when it has been in the market long enough to have allowed Cisco to fix enough bugs (Yuck!) and Cisco is comfortable with just about anyone using the software.
Feature sets don't change as often as version numbers. You select which feature set you want based on what you want to run on your router. For example, do you want to run just the Internet Protocol (IP), or do you want to run IP, Novell's Internetwork Packet eXchange (IPX), and DECnet? From your requirements, you should pick the feature that includes all of the features that you need for placing the router into operation in your network.
There are many models of routers that run IOS. They vary from the very inexpensive, low-end models to the extremely expensive, high-end models. You determine your model by the router's purpose and its cost. If you need a router for your network backbone, you would probably select from one of Cisco high-end router series: 7000, 7200, 7500, or 12000. These series of routers are meant to be fast and reliable, and you can put many interfaces into them. If your new router is to be used to connect office LAN's or WAN's to your backbone, then you would select from one of the access-type router series: 1000, 1600, 2500, 2600, 3600, 3800, or 4000. In the examples done in this book, we'll be using Cisco 2520's, one of the 2500-series of routers. (There are over 20 models in the 2500 series.) I'm not going to discuss money here; that changes too fast, but usually the lower the model number, the lower its cost.
The nice thing about IOS configuration is that its commands are consistent across the entire IOS-based router line. This means that you have to learn only one command interface. This interface happens to be a command line interface so it looks the same whether you are talking to a router through a console port, a modem, or a telnet connection.
No matter what type of LAN or WAN interface you want, you can usually find it on at least one of the Cisco router models. After all, a router wouldn't be much good without interfaces, and one of the major components of IOS configuration is interface configuration. Some of the physical interface types that are available on the Cisco routers are as follows:
Please don't assume the introductory material in this section is definitive since things change rapidly and I'm not a sales person or a marketing person. I just want to give you some idea of the flexibility and depth of the Cisco router product line. If you really want to get more details about IOS versions, IOS feature sets, and router models, you can either check Cisco's web site, Cisco Connection Online (CCO), at http://www.cisco.com or call your local Cisco sales representative.
IOS configuration principles are consistent across versions, feature sets, and router models. We're going to cover those principles so let's get started.