Ccna10cd

GLOBAL KNOWLEDGE NETWORKä CERTIFICATION PRESS
Cisco Certified Network Associate Exam
Router and Switching Study Guide

Chapter 10

Wide Area Networking

BOOK TABLE OF CONTENTS

SELF TEST

RETURN TO HOME PAGE

Certification Objectives *

Configuring ISDN BRI *

Configuring ISDN PRI *

Virtual Circuits *

Error Correction *

Logical Interfaces *

Configuring Frame Relay on a Cisco Router *

From the Classroom *

Addresses and Connection Identifiers *

PPP Configurations *

Configuring Multilink PPP *

Certification Objectives

Configuring ISDN

Configuring X.25

Configuring Frame Relay

Configuring ATM

Configuring PPP and Multilink PPP

These days, almost any business of significant size needs, or already has, some form of a WAN in its infrastructure. Unfortunately, because of many companies’ poor preparation and limited knowledge of the subject, many such WANs are not functioning to their potential. This chapter’s goals are to outline the specifics of the WAN technologies currently available, and how to configure each one.

Configuring ISDN

Today’s current analog modem connections for Internet connectivity have two major drawbacks. First, whenever a line is used for data transmissions, a voice line is taken up. This can obviously be a serious problem when a small business might only have one line! Second, current technology only allows a throughput of 56 kbps, and this is possible only if the quality of both the connections and the phone system is nearly perfect. The solution to this problem has been around for several years, but is only now becoming popular: ISDN. ISDN is the digital equivalent to the analog phone line. The phone companies developed it in an effort to deliver not only voice but also high-speed data at a reasonable price, and across the already installed copper wires.

ISDN lines are composed of multiple 64Kb bearer channels (B channels) and one data channel of either 16Kb or 64Kb (D channel). The purpose of each might sound reversed, but the bearer channels are what transmit the data or carry the voice signals, and the data channel is what is used for call signaling. ISDN is ordered as either {Answer to Self Test Question #21} Basic Rate Interface (BRI) or Primary Rate Interface (PRI). It is comprised of two B channels along with one 16Kb D channel, enabling a possible throughput of 128Kb.

{Answer to Self Test Question #1}BRI is typically configured as a type of dial-on-demand routing (DDR). A DDR link looks the same to IP, or to any other network protocol for that matter—as a serial link that is only there when in use. Configuration of IP on a DDR is pretty straightforward, but it can be fairly complex to determine how to specify the mapping of IP addresses to phone numbers, and when and how to connect or disconnect the links.

Configuring ISDN BRI

In order for a call to be placed across an ISDN network, there needs to be network-wide configuration information. {Answer to Self Test Question #26}ISDN uses directory numbers and service profile identifiers (SPIDs). The directory number is simply a telephone number you will use when you call. The SPID is a number the telephone company uses to identify the equipment on your ISDN connection. Another thing you need to know is the switch type used by your ISDN provider. {Answer to Self Test Question #22}In the United States, the switch type is most likely either an AT&T 5ess or 4ess, or a Northern Telecom DMS-100. A diagram of a basic ISDN network is shown in Figure 10-1.

Figure 1 An ISDN network{Answer to Self Test Question #2}

Let’s look at the configuration of the sample found in Figure 10-1.

Hostname router1

Isdn switch-type basic-dms 100

interface BRI0

ip address 10.10.10.1 255.255.255.0

isdn spid1 1234567890 5551234

encapsulation ppp

ppp authentication chap

dialer-group 1

dialer idle-timeout 300

dialer map ip 10.10.10.2 name Router2 speed 56 broadcast 2465551212

dialer hold-queue 5

dialer load-threshold 100

Following the host name of the router is the global command specifying the type of switch used on the phone company’s end. Then, after entering the configuration of the interface, which is BRI 0 on Router 1, you set the IP address (if applicable) and the SPID and directory numbers assigned to you by the phone company. The two lines referencing PPP will be discussed later in the chapter.

The last five lines all deal with the dialer’s settings. {Answer to Self Test Question #23}The DIALER-GROUP command adds the interface to the group listed, stating it will respond to interesting packets. Interesting traffic is traffic that, when routed to the interface that has been placed in the dialer-group, will cause the interface to dial the remote router and make a connection. Traffic routed to the interface that does not match the dialer-list will be dropped if the circuit is down, and will not cause the circuit to be activated. When the period of time defined by the dialer idle-timeout has elapsed without interesting traffic (there might be other traffic on the line once it is up that is not interesting), the dialer will terminate the connection.

The idle-timeout references, in seconds, how long of a period the system will wait where there is no activity before disconnecting. The next line actually does the mapping of the IP address of the remote router to its number. The hold-queue specifies the number of interesting packets the router will hold in its queue while a connection is being established. The dialer load-threshold configures the bandwidth maximum load before another dial-on-demand call is placed. While this particular line has no real significance in this scenario (since we only have one dial-up setting), I wanted to show you an example, since you will encounter it in most multi-modem configurations.

Configuring ISDN PRI

{Answer to Self Test Question #3}Implementation of ISDN PRI loses the benefit of using existing copper. {Answer to Self Test Question #24}It is delivered via a T-1 circuit and is comprised of 23 B channels and one 64Kb D channel. One thing to note is that when a T-1 is used for any type of channelized service, 8 kbps is lost to the channelization process. Since a T-1 provides 1.544 Mbps, its bandwidth is reduced to 1.536 Mbps.

A common implementation of ISDN PRI would be to provide 23 ports for a company’s remote workers to dial in. ISDN PRI has two features that combine to make using it an attractive remote-access (RAS) solution. One is that an ISDN line uses separate phone numbers for each channel, just as the analog lines in your house each have their own number. The other is that an ISDN PRI interface can be terminated into a RJ-45 connector, which can be directly connected to a Cisco access router or an ISDN PRI interface on a 7000. What this means is that you can introduce a great RAS solution to your company without having to mess with multiple lines, cables, and modems.

Configuring X.25

{Answer to Self Test Question #25}X.25 is similar to Frame Relay (which will be discussed in the next section) in that it is a packet-switched technology that typically operates as permanent virtual circuit (PVC). Since data on a packet-switched network is capable of following any available circuit path, it is almost always drawn as clouds in graphical representations like the one in Figure 10-2. Being configured as a PVC means that all data entering the cloud at point A is automatically forwarded to point B.

Figure 2 Basic X.25 configuration

X.25 was introduced at a time when wide-area network links, traveling through the public switched network, were primarily analog lines producing errors and poor transmissions. X.25 sought to remedy this through built-in error correction and flow control. The trade-off for this reliability is performance. With all the acknowledgments, buffering, and retransmission that happens within X.25, latency becomes an issue. In the grander scheme of things, for protocols that provide their own error detection and correction, such as TCP, it is a poor performer. {Answer to Self Test Question #4}

X.25 encompasses the first three layers of the OSI model. On each end of an X.25 connection, along the physical layer of the OSI model, is a data terminal equipment (DTE) device and a data circuit-terminating equipment (DCE) device. ATM connections occur between these devices on logical circuits. Circuits exist as one of two types: SVC or PVC. SVCs (switched virtual circuits) are a lot like telephone calls—a connection is established, data is transferred, and then the connection is terminated. Each DTE on an X.25 network is given a unique address, which can be used much like a telephone number. PVCs (permanent virtual circuits) are closer to a leased line idea, in that their connection is always present. No call needs to take place, since the connection is already established.{Answer to Self Test Question #5}

To establish an SVC connection, the calling DTE sends a Call Request packet, which includes the address of the remote DTE to be contacted. The Call Request packet contains both the sender and the destination DTE addresses, along with some other configurable information. At this point, the called DTE can use this information to decide whether or not it wants to accept the call. If it decides to accept it, a Call Accepted packet is sent. If it decides not to accept it, a Clear Request packet is sent.

{Answer to Self Test Question #6}Once the DTE that placed the call receives the Call Accepted packet, the virtual circuit is established and data transmissions begin to occur. At any point in the communication process, either DTE can terminate the call by issuing a Clear Request packet to the other DTE. It then responds with a Clear Confirmation packet, and the call is ended. The destination for each packet sent is identified by means of the logical channel identifier (LCI) or logical channel number (LCN). This allows the PSN to route each packet to its intended DTE.

(This is a lot like DLCI numbers you will read about in the Frame Relay section of this chapter.) {Answer to Self Test Question #27}LCNs only have significance locally, so many of the same number can be found throughout a given X.25 network.

Devices throughout an X.25 network have unique address assigned to them, which are known as network user addresses (NUAs). These addresses conform to the X.121 recommendation for public data networks. {Answer to Self Test Question #28} This specification stipulates that 14 digits be assigned, with 12 being mandatory and 2 being optional. The first four digits are known as the Data Network Identification Code (DNIC). Out of these four digits, the first three signify the country and the fourth specifies the network within the country. The next eight digits are the national number, and the last two are the optional ones I mentioned. They are used for sub-addresses, and are assignable by the network user.

{Answer to Self Test Question #7}X.25 relies on the underlying strengths of the layer-2 HDLC LAPB protocol to transmit data across an X.25 network. HDLC LAPB is a very efficient protocol. A minimum amount of overhead is required to perform flow control, synchronization, and error recovery. If the connection is configured as full duplex, where data is flowing in both directions, the data frames themselves carry all the information required to provide data integrity.

X.25 makes use of something called frame windows. A frame window is used to send multiple frames before receiving confirmation that the first frame has been received. This means that data can continue to flow, unlike the other protocols where there may be latency because of the sending and receiving of integrity confirmation-type packets.

An X.25 packet makes up the data field of an LAPB frame. Additional flow control and windowing are provided for each logical channel at the X.25 level.

{Answer to Self Test Question #29}Maximum packet sizes vary from 64 bytes to 4096 bytes, with 128 bytes being the default on most networks. Both maximum packet size and packet-level windowing may be negotiated between DTEs on call setup.

I recently read an article where somebody accurately described X.25 as a data pump. What this means is that there has to be some higher-level protocol that is able to make sense of the bits. There are standards for allowing certain applications to make use of X.25. Among them is IBM’s QLLC protocol, which defines how SNA traffic can be carried over X.25 networks. Another is the asynchronous X.25 PAD.

Configuration of X.25 is pretty straightforward. Here is a sample configuration for Router 1 from Figure 10-2.

X25 routing

Interface ethernet0

Ip address 10.10.10.1 255.255.255.0

Interface serial0

No ip address

Encapsulation x25 dce

Clockrate 19200

X25 route 2468013579 interface serial0

Most commands listed above should be understandable to you now. The first command establishes the routing based on the NUA of the router. The next four commands specify the IP address, or lack thereof, for each interface on the router. The ENCAPSULATION command ends in DCE, thereby configuring the serial port to be a X.25 DCE device. The CLOCKRATE command lets you specify to the router the rate at which the data is going to be transferring. The last command tells the router to send all traffic destined for the listed NUA out through serial 0.{Answer to Self Test Question #8}

If you want to view the status of your X.25 connections, you can use the SHOW X25 command, followed by one of the arguments listed in Table 10-1. (for example, show x.25 map) You should try each of these to view the output and see how it can help you in the troubleshooting process.

Argument	Definition
MAP	Displays X.25 address maps
ROUTE	Displays the X.25 routing table
VC	Lists information about active SVCs and PVCs
REMOTE-RED	Displays one-to-one mapping of local and remote IP addresses

Table 1 Show X.25 Arguments {Answer to Self Test Question #9}

Configuring Frame Relay

Frame Relay is not a certain type of interface, rather it is an encapsulation method that operates at layer 2 and runs on top of nearly any serial interface. Frame Relay is a packet-switching technology that multiplexes multiple logical data streams onto one physical link. These data streams are called virtual circuits, and each is identified by something known as a data-link connection identifier. The acronym for this is DLCI, which is pronounced dell-see. {Answer to Self Test Question #30}This DLCI number is not used as a network-wide unique descriptive destination, like a MAC address is on a typical layer-2 protocol. It only has significance locally, and can change on each physical link. The number is used to identify which pipe leads to which specific layer-3 protocol address, such as IP. In the example in Figure 10-3, DLCI 4 leads to the IP address of serial 0 at the location of Router 2.

Figure 3 A Frame Relay network

In a Frame Relay cloud, other locations could, and most likely do, use the same DLCI numbers, since they are only locally significant. This really puts a limit on its functionality, and it has in fact been antiquated by use of Local Management Interface (LMI) extensions. {Answer to Self Test Question #31} LMI provides Frame Relay with many new features, with one of the most important being the use of inverse ARP to automatically determine the protocol address of the device on the remote end of a DLCI. You can just accept this as being a good thing for now, and later in this section we will see how much this helps us in configurations! LMI also adds flow control functionality to Frame Relay, as well as the capability to multicast.

Virtual circuits are statistically multiplexed, meaning bandwidth is dynamically assigned as needed per circuit. This technique is the opposite of another type of switching technology called time-division multiplexing (TDM), where each and every channel is allocated a fixed amount of bandwidth over a given time. The disadvantage to TDM is that each channel is given its specified bandwidth, whether it takes advantage of it or not. Let’s look at the following example to understand not only the rationale but also the implementation of dynamic allocation. {Answer to Self Test Question #10}

If a company requires a leased line between two locations, the amount of bandwidth they need will fluctuate constantly throughout the day. If they purchase an amount close to the average, there will be occasions when they max out their line and network traffic slows to a crawl. Yet they don’t want to spend the extra money every month for bandwidth they rarely use. However, if multiple companies are in this situation and they share a line that provides more than adequate bandwidth for each of their averages collectively, the odds that all the companies involved will all require their maximum bandwidth simultaneously is statistically improbable. So bandwidth is dynamically assigned to each company where needed. This results in greater efficiency of bandwidth allocation collectively, which itself produces a reduction of costs to and from the phone company.

Many people are concerned when first presented with this option, thinking they might not get the bandwidth required just for normal, day-to-day operations. To prevent this from happening, Frame Relay provides something called the committed information rate (CIR). {Answer to Self Test Question #44}What this does is guarantee the customer a fixed amount of throughput based upon what they deem acceptable. In other words, the customers set the minimum they require, and even though their line is shared, their available bandwidth will never drop below that amount.

Virtual Circuits

There are two types of virtual circuits that Frame Relay can create. This is exactly like what you learned in the X.25 section, but for clarity, we will re-explain it. The first type of circuit is a permanent virtual circuit (PVC) and the second is a switched virtual circuit (SVC). A PVC is manually created by an administrator with a source and destination, and operates very much like a leased line. As its name implies, it is a permanent connection and remains until it is manually removed. An SVC on the other hand, is dynamically created by software through a call setup procedure. This is similar to the process by which two people operate a telephone. When communication is required, a call is placed, and it is disconnected when the transmission has ended.

Since these circuits are multiplexed on each link, Frame Relay can operate in either a point-to-point or a multipoint mode. Point-to-point operation is just like that of a leased line. Communication exists across a single virtual circuit between two interfaces at two destinations. In multipoint mode, an interface can communicate with one of multiple destinations across one of multiple circuits. While an interface has the capability to communicate with each remote location, each communication is performed as an independent, point-to-point session. In other words, a broadcast packet cannot be sent without recreating it onto each individual circuit.

Error Correction

Frame Relay also has error correction built into it, but not nearly to the extent that X.25 has. Similar to the cyclic redundancy check (CRC) in an Ethernet network, Frame Relay uses a Frame Check Sequence (FCS) and it is appended to the end of each frame passed. {Answer to Self Test Question #32} When a station receives a frame, it computes a new FCS on the data portion and compares it to the FCS that was in the frame. If they are different, it drops the packet without notifying the sending station. While this may sound bad, it is in fact a good thing. Because of this technique, Frame Relay is faster at transferring data than X.25, because no time is lost in the overhead of having to process error checking or having to re-send information. Instead, Frame Relay relies on the next layer that is communicating over it to handle error recovery, since most level-3 protocols do. If the protocol that is running over Frame Relay is connection oriented, such as the TCP half of TCP/IP, there are no problems, since it will handle it’s own error recovery and flow control. But if the protocol is connectionless, like that of UDP, the application that is implementing it must be specifically coded for self-recovery.

Logical Interfaces

Frame Relay, like ATM, allows multiple logical network interfaces to be bound to each physical interface. Each of these logical interfaces can have its own IP address, subnet mask, and entry in the routing tables, and each will have a different subnetwork ID associated with it. Each of these logical interfaces can run on one or even multiple virtual circuits, but each virtual circuit typically only has one logical interface associated with it. To assign subnetwork IDs to each interface, Cisco IOS uses what are called sub-interfaces. Sub-interfaces allow you to split up each of your physical interfaces into multiple logical ports. The naming convention used for these ports is simply the name of the interface being addressed, then a period, followed by the number of the sub-interface. From Router 3 in the Frame Relay example in Figure 10-4, serial 0.2 is the second sub-interface on serial 1. {Answer to Self Test Question #11}

Figure 4 Frame Relay allows multiple logical network interfaces on each physical interface
Configuring Frame Relay on a Cisco Router

Configuring Frame Relay on a Cisco router includes the mapping of IP addresses to DLCIs and telling the router which virtual circuits are connected. The way this is done is the same whether it is operating in point-to-point or in multipoint mode. The only difference is, whatever you do to a point-to-point interface has to be repeated for each logical circuit on a multipoint. Point-to-point and multipoint connections operate in either explicit or implicit mode. In explicit mode, a map of the remote IP address to the DLCI is made manually. In implicit mode, an assumption is made that the router on the other end of the circuit matches the IP address it wants to send the packet to. You will probably recognize this concept as being similar to IP unnumbered, and in fact point-to-point Frame Relay can be configured in this way! {Answer to Self Test Question #34}To configure Router 3 in Figure 10-4 for explicit mode, you would enter the following:

interface serial0

encapsulation Frame Relay {ietf}

interface serial 0.1 point-to-point

IP address 10.10.10.3 255.255.255.0

Frame Relay map 10.10.10.1 7 broadcast

The first command configures the interface for Frame Relay encapsulation. {Answer to Self Test Question #33}The IETF at the end optionally changes the encapsulation method from Cisco’s own method to comply with the IETF standard. This is used in situations where the router at the other end is a non-Cisco product. The INTERFACE command creates a point-to-point sub-interface and the next line maps an IP address to it. The last line is where the explicit configuration is made. It states that at the end of the DLCI 7 pipe is the destination IP address of 10.10.10.1. The BROADCAST argument tells the router that broadcasts such as router updates should be sent across this PVC. Of course, to be able to send router updates both ways, Router 1 would have to be configured to also allow broadcast messages.

The next example assumes an implicit configuration is desired, and now we get to see the great LMI added feature of reverse ARP that we promised earlier!

interface serial0

encapsulation Frame Relay {ietf}

Frame Relay lmi-type ansi

interface serial0.1 point-to-point

IP address 10.10.10.3 255.255.255.0

Frame Relay interface-dlci 7 broadcast

A lot of this looks the same, but the Frame Relay lmi-type is new. {Answer to Self Test Question #35}This command enables LMI extensions, and specifies which of these three standards it is to implement: ansi, q933a, or the default of Cisco. The FRAME RELAY command in the last line associates DLCI 7 to the sub-interface, but why is the IP address of the remote interface not listed?

By simply telling the router of the DLCIs in use, the router will use inverse ARP to build a table of the IP address of the (sub-)interface at the end of the PVC, matched with each respective DLCI. The use of inverse-arp rather than explicit configuration would obviously save time and simplify the setup and management if you had multiple sites and each site had multiple PVCs. {Answer to Self Test Question #12}

From the Classroom

Addresses and Connection Identifiers

At this point in the ICRC course it would be time to do a Frame Relay lab exercise so you could apply what you have learned about this popular WAN service. This is also where I would learn how well or how badly I had gotten the key points across (or who was playing solitaire while I was explaining about DLCIs). The thing that causes the most trouble for students in configuring Frame Relay is understanding the DLCI and what it actually represents.

As the name implies, the Data-Link Connection Identifier is a number used to identify a particular permanent virtual circuit (PVC) connection to a Frame Relay switch. It is not an address. The DLCI is often said to be "locally significant" between the switch and the router. What that means to you when you configure your router for Frame Relay service is that you don’t need to worry about the DLCI at the other end of the PVC. It may be the same, or it may be different; it doesn’t matter. The most common mistake students make in the Frame Relay lab is to try to figure out what the DLCI is at the destination and try to put the remote DLCI into their local router configuration. A look at what happens inside the Frame Relay "cloud" might be helpful in understanding why that won’t work.

The Frame Relay service provider will give you an access line (a T-1, for example) that connects your site to a switch providing an access point into the provider’s "cloud." You may have several destinations (let’s say Denver, Chicago, and Atlanta) to be reached via the same access line. This capability is what has made Frame Relay such a popular service. The provider will configure a PVC through his network to each of these locations. The provider will assign a different DLCI for each PVC. If you have configured your router correctly, the DLCI for the PVC to Atlanta will be placed inside the frame header for all the traffic destined for Atlanta as it leaves your router. Likewise, the DLCIs for Chicago and Denver will be placed inside the frame headers encapsulating the traffic destined for those locations. Traffic leaving your router for each of three destinations is sharing the same interface and access line, and the only way the provider’s switch can sort out one destination’s traffic from another’s is by inspecting the DLCI in the frame header. The DLCI tells the provider’s switch which PVC to use for that traffic. The switch your router is connected to doesn’t know about DLCIs anywhere else in the network. If you try to configure the remote DLCI on your local router, the switch wouldn’t know what to do with that traffic, because it only knows about its locally configured DLCIs.

Once the traffic gets inside the cloud it may get a totally different identifier, or it may even be converted to ATM cells. In any case, the DLCI it had when it entered the first switch is no longer being used because WAN switches have their own conventions for identifying PVCs. When the traffic gets to the destination switch, though, it will be converted back to Frame Relay (if necessary) for transmission to the destination router, and it will be given a DLCI that identifies the same PVC, but this time between the destination router and its access switch. Even though this DLCI identifies the same PVC, the DLCI may be a different number at this destination end of the circuit.

Just to confuse the issue a little more, some providers now offer a feature called global addressing for their frame relay service. Don’t be fooled by this terminology. With global addressing, the provider will establish consistent DLCIs for your organization across his network for each of your PVC destinations. For example, let’s say you have PVCs between Houston and Atlanta, between Chicago and Atlanta, and between Denver and Atlanta. The DLCIs at Houston, Chicago and Denver that identify PVCs with an Atlanta destination will all use exactly the same number. This makes it appear that the DLCI has become an address, because every router in your network with a PVC to Atlanta will use the same DLCI number to refer to that PVC. The DLCIs are still locally significant, and the switches and routers still use them in the same way when you have global addressing. The only difference is in the way the numbers have been assigned.

—By Pamela Forsyth, CCIE, CCSI, CNX
Configuring ATM

At first glance ATM seems very close to Frame Relay. Using switching and multiplexing technologies, virtual circuits, and dynamic bandwidth allocation, it is obvious that ATM was at least based on the foundations of Frame Relay. Where Frame Relay ends as an excellent WAN technology, ATM continues to the LAN. ATM blurs the lines between LAN and WAN technologies, creating for the first time a viable all-in-one solution.

{Answer to Self Test Question #36}One key difference between ATM and Frame Relay is the guarantee of delivery. Earlier we discussed how Frame Relay was a layer-2 technology that relied on the encapsulated layer-3 protocol for error-recovery. ATM differs in that, depending on the transmission, it has the capability to provide a guaranteed delivery at a specific rate.

ATM’s packet sizes are created at a fixed length, instead of varying like Frame Relay and X.25. As shown in Figure 10-5, the ATM cell is 53 bytes long and is referred to as a cell. A five-byte header contains the address information and other fields of information used to route the cell through the network. After the header comes a 48-byte information field called a payload. Because of this fixed length, ATM can predict and control the number of packets, to control bandwidth utilization. {Answer to Self Test Question #13}

Figure 5 The ATM cell

Having a cell of a fixed length also means that buffers can be designed at a set length, thereby allowing hardware switching. Using switching technology in hardware rather than in software tables helps minimize latency for time-critical data such as video and sound.

One of the reasons ATM is so fast is as a result of its use of virtual channels and virtual paths to route traffic through the network. By implementing virtual channel connections (VCC), the routes to be used by the ATM routing device are determined before data is even transferred. Using this method, the transfer of data does not require complex routing decisions to be made in real time through software-based routing tables. Routing decisions are made in the hardware, thereby minimizing latency in data transfer.

As I just explained, the VCC is an identification for the path that data will travel over. To be more specific the VCC is an index determined by the combination of a one-byte virtual path identifier (VPI) and a two-byte virtual channel identifier (VCI). The VPI can be thought of as a larger group that contains either multiple VCIs or just one single VCI. In the vast majority of the cases, the VPI will be equal to 0, which is its default. The VCI identifies the individual circuit within the larger VPI. Figure 10-6 gives a visual representation of how the VPI and VCI could relate between two switches.{Answer to Self Test Question #14}

Figure 6 Virtual path identifiers contain virtual channel identifiers

The VPI and VCI numbers in ATM are similar to the DLCI number found in Frame Relay, in that they only have relevance locally. In this case "locally" refers to a segment, which can be either between a host and a switch, or between two switches. Even though two switches might recognize the VCC by different numbers, it is still the same circuit.

Also just like in Frame Relay, virtual circuits can be categorized into two groups: permanent virtual circuits and switched virtual circuits. A permanent virtual circuit is a connection between end points that is not dynamically established or removed. If you’ll recall, PVC connections are manually implemented and manually released. Implementation in an ATM network is typically found at the WAN level. A switched virtual circuit is a connection that is dynamically established and released. It is most often found at the LAN level, all the way to the desktop.

ATM is often referred to as a shared-media LAN. The one characteristic of shared-media LANs that is fairly obvious is that with each user you add, less is available for everybody else. Hosts must contend for access to the transmission medium, since in a shared-media LAN, the network is available to only one user at a time.

ATM LANs can operate over several different types of media. Using a special piece of hardware, ATM can run at 155 Mbps over Category 5 twisted pair. And while this is the most widely adapted configuration to the desktop, ATM also has the capability to run at 25 Mbps over two pairs of Category 3 or 4 cable. {Answer to Self Test Question #39}Finally, for higher ATM speeds and distances of more than 100 meters, fiber-optic cable is required. Over fiber, ATM can run up to 622 Mbps.{Answer to Self Test Question #15}

Implementation of ATM can occur by one of two methods. The first method applies specifically to the routing of IP over ATM. The Internet Engineering Task Force (IETF) has adopted this support and defined a Request For Comment (RFC) 1577 on Classical IP and ARP over ATM. Cisco supports RFC 1577 in the Cisco 4000, 7000, and 7500 router families.

The second method for data transfer across an ATM network is through something called LANE technology. LANE is a layer-2 bridging protocol that takes an ATM network, which is connection-oriented, and modifies it to look and behave like a shared connectionless Ethernet or Token Ring LAN segment. {Answer to Self Test Question #40}Since it acts as a layer-2 service, LANE can handle not only routable protocols such as TCP/IP and IPX, but also non-routable protocols such as NetBIOS and SNA. {Answer to Self Test Question #16}

Now that you have a pretty good background on ATM, here is a sample configuration.

Interface atm1

Ip address 10.10.10.1 255.255.255.0

Atm pvc 1 0 32 aal5mux ip

Atm pvc 2 0 45 aal5mux ip

Map-group samplegroup

Map-list samplegroup

Ip 10.10.10.5 atm-vc 1

Ip 10.10.10.6 atm-vc 2 broadcast

Each of the ATM PVC statements creates PVCs on the interface. The first number is the VCD, the second is the VPI, and the third is the VCI. {Answer to Self Test Question #41}One thing to note is that either the VPI or the VCI can be 0, but not both. The last argument lists the adaptation layer and encapsulation type. When aal5mux is chosen, the following statement must specify DecNET, IP, Novell, Vines, or XNS as the protocol being used.

The MAP-GROUP command assigns a map-list to the interface. The MAP-LIST command defines an arbitrary name to the subsequent group that will be defined. The list is defined with the IP statements, specifying the IP address to map to each VCD. {q38}If you will recall, ATM does not directly do broadcasts, but it can pseudo-broadcast by replicating each broadcast packet across each VC that is set up to receive them, with the broadcast argument at the end.

Configuring PPP and Multilink PPP

The Point-to-Point Protocol (PPP) is described in RFCs 1661 and 1332. What PPP does is encapsulate network layer protocol information over point-to-point links. PPP can be configured on the following types of physical interfaces.

ISDN
Asynchronous serial
Synchronous serial
HSSI

Since PPP encapsulation is bound to physical interfaces, PPP can also be set up on calls placed by the dialer interfaces that use the physical interfaces. The current implementation of PPP supports:

Authentication using CHAP or PAP

The binding of multiple protocols over one link

Dynamic address assignment

Both synchronous and asynchronous communication. {Answer to Self Test Question #18}

We will look at each of these in detail later in this section. We will not address PPP encapsulation across synchronous connections directly, since the configuration of PPP itself is the same. You should be able to tell which parts only apply to async connections.

PPP Configurations

The first step in establishing a PPP connection is the authentication process. Authentication using Password Authentication Protocol (PAP) is a step better than nothing, but is in no way a secure method. In a PAP negotiation, the remote router attempting to connect to the local router or access server is required to send an authentication request. If the username and password specified are correct, the Cisco IOS software sends an authentication acknowledgment.

The way Challenge Handshake Authentication Protocol (CHAP) works is, when a remote device such as a router wishes to make a connection with a local device, the local router or access server sends a CHAP packet to the remote device. What this CHAP packet does is request or "challenge" the remote device to respond. The challenge packet consists of an ID, a random number, and the host name of the local router. The required response contains an encrypted version of the ID, a one-way encrypted password, the random number, and either the host name of the remote device or the name of the user on the remote device.

{Answer to Self Test Question #45}After the local device receives the response, it verifies the secret by performing the same one-way encryption on its password and compares the results. If they match, it then looks up the remote device’s host name or username. Since the password is transmitted by this encrypted method, it is never sent in clear text. These CHAP transactions only occur when the link is first established. The local router or access server does not initiate a request for a password for the duration of the call. The following listing outlines the beginning steps to establishing a PPP connection.

Interface serial0

Encapsulation PPP

PPP authentication {CHAP | PAP} {if-needed}

The majority of this first part is pretty self-explanatory. You first enter which interface you are binding the information to, which in this case is serial 0. You then specify to encapsulate PPP, followed by the authentication method. The IF-NEEDED argument is used in TACACS and XTACACS configurations, whereby PPP or CHAP authentication is not performed if the user has already provided authentication. It would then be followed with the name of a list of TACACS authentication methods to use. The lists are created with the aaa authentication ppp command. One thing to note is that the if-needed option is only available on asynchronous interfaces.{Answer to Self Test Question #19}

The next step in configuring PPP is defining which network protocol to encapsulate. Generally you have the choice of IPCP and IPXCP. As you can probably guess from their names, IPCP is for IP traffic and IPXCP is for IPX traffic. It is possible, however, to provide support for both simultaneously across the same session. Each layer-3 protocol negotiates the specifics for its own communication independently. This process is the same as with the other WAN technologies we’ve already discussed. For instance, in an IP environment you can go with either IP classless or static assigned addresses. IP addresses can also be assigned dynamically in the PPP authentication process.

As you probably already know, either hardware or software can be used to control the flow for asynchronous communications. Hardware flow relies on pin signaling like the data set ready (DSR) or the data terminal ready (DTR) pins. Software flow control uses specific characters sent during the transmission to signal starts and stops. The problem with this is that there is always the chance that a string to be transmitted across could be the same as that of a flow control command. {Answer to Self Test Question #46}This is where asynchronous control character maps (ACCMs) come into play. An ACCM tells the port to ignore specified control characters within the data stream. In order for an asynchronous port to know to ignore XON/XOFF (the software control), the hexadecimal number A0000 must be passed. If, however, the router at the other end of the connection does not support ACCM negotiation, the port will be forced to use FFFFFFFF. If this is your situation, you can easily set the ACCM manually with the following command:

Ppp accm match 000a0000

Another configuration issue to address is the DTE rate. The DTE rate is the speed at which a router sends information to the modem. The modem, in turn, must decompress the data it receives before it can be sent to the router. If the amount of decompressed data is larger than the DTE rate, then data will bottleneck and slow down. {Answer to Self Test Question #42}For example, let’s say we set the DTE to 38400.

Line 1

Rxpseed 38400

Txspeed 38400

Now, if our connecting modems are capable of establishing 28.8 connections, data will flow to the modem at 38400, but the modem will not allow more than 28.8 to flow across the line. This is where compression comes in. Using compression schemes such as V.42bis, throughput can be increased by a factor of four.

In order to control the amount of buffered packets held before the router starts dropping them, Cisco IOS has the hold-queue command. {Answer to Self Test Question #47}The following commands increase the amount to 100 for both incoming and outgoing.

Hold-queue 100 in

Hold-queue 100 out

PPP also supports link quality monitoring (LQM). LQM will do just as it says. It monitors the quality of the link, and if the quality drops below a pre-configured percentage, the router shuts down the link. LQM is enabled for both the incoming and outgoing directions. The quality is calculated by comparing the total number of bytes sent with the total number of bytes received by each device each direction. When LQM is enabled link quality reports (LQRs) are sent every keepalive period, and in fact are sent instead of the normal keepalives. {Answer to Self Test Question #43}To setup LQM, you enter the command PPP QUALITY percentage, where percentage is a number from 1 to 100 that quality should not drop below.

One thing you need to decide on is whether or not to use the Cisco IOS compression. Compression reduces the size of a PPP frame via lossless data compression. The compression algorithm typically used is the RAND predictor algorithm, which uses a compression dictionary to predict the next character in the frame. In the Cisco 7000 series only, another compression alternative is available: hardware and distributed compression. Whether or not this it is available in your 7000 depends on the interface processor and compression service adapter hardware installed in the router.

It is Cisco’s recommendation that you disable software compression if the router CPU load exceeds 65 percent. To view the CPU load, use the command Show process cpu in EXEC mode.

Also, if the majority of your traffic is already compressed files, do not use compression. Compression is enabled with the PPP predictor command:

PPP predictor {compress | stac}

Another form of compression is TCP header compression, also known as Van Jacobsen header compression. You will almost always use this form of compression in asynchronous modes. Obviously, TCP headers aren’t a large percentage of the packets sent, but in asynchronous communication, every little bit of bandwidth savings helps.

Ip tcp header-compression on

Configuring Multilink PPP

Multilink Point-to-Point Protocol (PPP), according to RFC 1717, provides the following functionality:

Load balancing over multiple WAN links
Packet fragmentation with proper sequencing in the resassembly
Load calculations on both inbound and outbound traffic

We will be addressing Cisco’s conformation to the packet fragmentation and sequencing division of this standard. What Multilink PPP (MLP) does is allow packets to be fragmented and sent at the same time over multiple point-to-point links to the same destination. The multiple links needed are established in response to a dialer load threshold that is defined. The load can be calculated on either the inbound or outbound traffic needed between the sites. In other words, Multilink PPP provides the capability of splitting and recombining packets to a single destination across one logical pipe. This concept is similar to the combing of VCIs to form VPIs in ATM networks. Multilink PPP is able to provide bandwidth on demand and reduce latency across your WAN.

Multilink PPP can work over single or multiple interfaces of the following types:

Asynchronous serial interfaces
Basic Rate Interfaces (BRIs)
Primary Rate Interfaces (PRIs)

Each interface must be configured to support both dial-on-demand routing (DDR) and PPP encapsulation. A sample configuration of the multilink portion of a PPP connection is as follows:

Interface dialer 100

No ip address

Encapsulation ppp

Dialer in-band

Dialer load-threshold load [inbound | outbound | either]

Ppp multilink

The first command assigns a number from 0 to 255 to the dialer rotary group. I’ve chosen not to use an IP address here, but this option is available if your situation requires it. The next two statements enable PPP encapsulation and the DDR functionality for the interface. The dialer load-threshold command sets the maximum threshold before another call is placed. This number is between 1 and 255, and is required. The last argument is optional, with the EITHER argument meaning that the calculated load is the maximum of the inbound and the outbound load. Finally, the last command enables the multilink functionality.

I need an all-in-one solution for my new LAN/WAN. Cost isn’t really an issue.	ATM would be an excellent solution, since it can travel over fiber and twisted pair.
My company needs a new inexpensive solution for Internet connectivity. We have found that we need more bandwidth than an ISDN BRI interface can provide.	Try implementing Multilink PPP over two or more ISDN lines.
The average bandwidth required by my company has increased over the past six months. We have a shared T-1 line, and don’t have the extra money to get our own dedicated.	Contact your ISP and have your committed information rate (CIR) increased to a little above your average.
We want to run classical IP over ATM. Which Cisco router would be best for a medium-sized business?	Classical IP is based on RFC 1577 and is supported in the 4000, 7000, and 7500 families. Depending on your budget, any of the three would be an excellent choice.
I need a secure method of connection and authentication for my point-to-point site. I am worried about somebody "listening" in on the line and getting access passwords. What should I use?	Use PPP encapsulation across the line with CHAP authentication. CHAP is much more secure than PAP, since passwords are never sent.
We are using software-based flow control, but we are having problems receiving the data correctly. What should we look at for possible problems?	Some of your data might contain the strings that are used for signaling a DSR or a DTR. Either implementation of hardware flow control or ACCMs would be a good place to start.

Certification Summary

ISDN is the digital equivalent to the analog phone line. It is comprised of multiple 64Kb bearer (B) channels and one data (D) channel that is either 16Kb or 64Kb in length. ISDN is ordered as either Basic Rate Interface (BRI) or Primary Rate Interface (PRI). ISDN BRI is typically configured as dial-on-demand routing (DDR) and ISDN PRI is typically used for RAS, since it can provide 23 ports for dial-in.

X.25 is a packet-switched technology that operates as permanent virtual circuit. X.25 has built-in error correction and flow control, thereby providing a guaranteed delivery. X.25 relies on the layer-2 protocol HDLC LAPB to transmit the data across the network. It compensates for latency in the network by using windowing techniques.

Frame Relay is an encapsulation method that operates at layer 2 on top of practically any serial interface. Frame Relay is a packet-switching technology that multiplexes multiple logical data streams onto one physical link. It uses DLCI numbers as a local reference for which pipe leads to which remote IP address. For ease of configuration, LMI extensions provide inverse ARP to automatically determine the protocol address of the remote device.

ATM is a lot like Frame Relay except that it doesn’t provide the guaranteed delivery. ATM’s packets are 53 bytes in length, and are routed through something called virtual channel connections (VCC). ATM can operate at 622 Mbps over fiber, 155 Mbps over Category 5, and 25 Mbps over two pairs of Category 3 or 4 cable.

PPP encapsulates network layer protocol over point-to-point links. Authentication can be performed with either CHAP or PAP. CHAP authentication is based on a random number and a password, thereby making it the more secure method. PPP also supports link quality monitoring (LQM) to monitor the quality of the link. Multilink PPP allows packets to be fragmented across multiple connections, and reassembles them into the proper sequence at the remote end.

Two-Minute Drill

ISDN is the digital equivalent to the analog phone line. The phone companies developed it in an effort to deliver not only voice but also high-speed data at a reasonable price, and across the already installed copper wires.
ISDN is ordered as either Basic Rate Interface (BRI) or Primary Rate Interface (PRI).
BRI is typically configured as a type of dial-on-demand routing (DDR). A DDR link looks the same to IP, or to any other network protocol for that matter—as a serial link that is only there when in use.
Implementation of ISDN PRI loses the benefit of using existing copper. It is delivered via a T-1 circuit and is comprised of 23 B channels and one 64Kb D channel. ISDN uses directory numbers and service profile identifiers (SPIDs).
X.25 is similar to Frame Relay in that it is a packet-switched technology that typically operates as permanent virtual circuit (PVC).
X.25 encompasses the first three layers of the OSI model. On each end of an X.25 connection, along the physical layer of the OSI model, is a data terminal equipment (DTE) device and a data circuit-terminating equipment (DCE) device.
Circuits exist as one of two types: SVC or PVC.
SVCs (switched virtual circuits) are a lot like telephone calls—a connection is established, data is transferred, and then the connection is terminated.
PVCs (permanent virtual circuits) are closer to a leased line idea, in that their connection is always present.
If you want to view the status of your X.25 connections, you can use the SHOW X25 command.
Frame Relay is not a certain type of interface, rather it is an encapsulation method that operates at layer 2 and runs on top of nearly any serial interface.
LMI provides Frame Relay with many new features, with one of the most important being the use of inverse ARP to automatically determine the protocol address of the device on the remote end of a DLCI.
Configuring Frame Relay on a Cisco router includes the mapping of IP addresses to DLCIs and telling the router which virtual circuits are connected.
ATM blurs the lines between LAN and WAN technologies, creating for the first time a viable all-in-one solution.
One key difference between ATM and Frame Relay is the guarantee of delivery.
One of the reasons ATM is so fast is as a result of its use of virtual channels and virtual paths to route traffic through the network.
The VPI and VCI numbers in ATM are similar to the DLCI number found in Frame Relay, in that they only have relevance locally.
ATM LANs can operate over several different types of media.
PPP encapsulates network layer protocol information over point-to-point links.
The first step in establishing a PPP connection is the authentication process. Authentication using Password Authentication Protocol (PAP) is a step better than nothing, but is in no way a secure method.
Multilink Point-to-Point Protocol (PPP), according to RFC 1717, provides the following functionality:
Load balancing over multiple WAN links
Packet fragmentation with proper sequencing in the resassembly
Load calculations on both inbound and outbound traffic

Self Test

What does DDR stands for?

Dial-up direct request
Direct dial relocation number
Dial-on-demand request
Dial-on-demand routing

CCNA Routing and Switching Study Guide: Self Test for Exam 640-407