Dial Configuration Guide, Cisco IOS Release 15.5M&T

Individual Remote PCs Using Analog Modems

ISPs can configure a single Cisco access servers to receive analog calls from remote PCs connected to modems, as shown in Figure 1. The point of presence (POP) at the ISP central site could also be a Cisco 2511 access server connected to external modems.

Network Topology

Figure 1 shows a small-scale dial-in scenario using modems.

Figure 1 Remote PC Using an Analog Modem to Dial In to a Cisco Access Server

Running Configuration for ISDN PRI

The following sample configuration runs on the Cisco access server, as shown in Figure 1, which enables remote analog users to dial in:

Some service providers use a remote TACACS+ or RADIUS security server in this dial-in scenario. The following example shows a TACACS+ entry that appears in the configuration file of a remote security server:

Running Configuration for Robbed-Bit Signaling

The following example shows a single Cisco access server configured to support remote client PCs dialing in with analog modems over traditional T1 lines. Digital ISDN calls do not transmit across these older types of channelized lines. The configuration assumes that the client can dial in and connect to the router in either terminal emulation mode (text only) or PPP packet mode.

Note The following configuration works only for analog modem calls. It includes no serial D-channel configuration (Serial 0:23 and Serial 1:23).

Individual PCs Using ISDN Terminal Adapters

ISPs can configure a single Cisco access server to receive digital multilink calls from remote PCs connected to terminal adapters, as shown in Figure 2. The POP at the central site of the ISP can be any Cisco router that supports ISDN PRI, such as the Cisco 4700-M router loaded with a channelized T1 PRI network module.

Network Topology

Figure 2 shows a small-scale dial-in scenario using terminal adapters.

Figure 2 Remote PC Using a Terminal Adapter to Dial In to a Cisco Access Server

To configure one Cisco access server to accept both incoming ISDN and analog calls from individual terminal adapters and modems, see the section “Mixture of ISDN and Analog Modem Calls” later in this chapter.

Terminal Adapter Configuration Example

The following example configures a Cisco access server to enable PCs fitted with internal or external terminal adapters to dial in to an IP network. The terminal adapter configuration is set up for asynchronous-to-synchronous PPP conversion. In some cases, PPP authentication must be set up for the Password Authentication Protocol (PAP). Some terminal adapters support only PAP authentication.

Mixture of ISDN and Analog Modem Calls

ISPs can configure a single Cisco access server to receive calls from a mixture of remote PCs connected to terminal adapters and modems, as shown in Figure 3.

Figure 3 Remote PCs Making Digital Calls and Analog Calls to a Cisco Access Server

Combination of Modem and ISDN Dial-In Configuration Example

The following example shows a combination of the modem and ISDN dial-in configurations. Using the bearer capability information element in the call setup packet, the incoming calls are labeled as data or voice. After the calls enter the access server, they are routed either to the serial configuration or to the modems and group asynchronous configuration.

Note This configuration assumes that only individual remote PCs are dialing in; no remote routers are dialing in. For a remote router dial-in configuration, see the chapter “Enterprise Dial Scenarios and Configurations” in this publication.

Scaling Considerations

Because of the significant increase in demand for Internet access, large POPs are required by many Telcos and ISPs. Internet access configurations can be set up to enable users who dial in with individual computers to make mixed ISDN multilink or modem connections using a stack of Cisco access servers that run Multichassis Multilink PPP (MMP).

You must consider scalability and call density issues when designing a large-scale dial-in POP. Because access servers have physical limitations, such as how many dial-in users can be supported on one device, you should consider the conditions and recommendations described in Table 1 .

Note Depending on the size of your POP requirement, you can replace the Cisco AS5200 access server with a Cisco AS5300, Cisco AS5800, or Cisco AccessPath. This hardware exchange provides higher call density performance and increases the number of ISDN PRI ports, channelized ports, and modem ports on each chassis.

How Stacking Works

Before you install and configure a stack of access servers, you should understand the basic concepts described in the following sections and how they work together in a large-scale dial-in solution:

• A Typical Multilink PPP Session
• Using Multichassis Multilink PPP
• Setting Up an Offload Server
• Using the Stack Group Bidding Protocol
• Using L2F

A Typical Multilink PPP Session

A basic multilink session is an ISDN connection between two routing devices, such as a Cisco 766 router and a Cisco AS5200 access server. Figure 4 shows a remote PC connecting to a Cisco 766 ISDN router, which in turn opens two B-channel connections at 128 kbps across an ISDN network. The Multilink PPP (MLP) session is brought up. The Cisco 766 router sends four packets across the network to the Cisco AS5200, which in turn reassembles the packets back into the correct order and sends them out the LAN port to the Internet.

Figure 4 A Typical Multilink PPP Session

Using Multichassis Multilink PPP

Note Effective with Cisco Release 12.4(11)T, the L2F protocol was removed in Cisco IOS software.

The dial solution becomes more complex when the scenario is scaled to include multiple multilink calls connecting across multiple chassis. Figure 5 shows a terminal adapter making a call in to the Cisco AS5200, labeled #1. However, only one of the access server’s 48 B channels is available to accept the call. The other channels are busy with calls. As a result, one of the terminal adapter’s two B channels is redirected to device #2. At this point, a multilink multichassis session is shared between two Cisco AS5200s that belong to the same stack group. Packet fragments A and C go to device #1. Packet fragments B and D go to device #2.

Because device #1 is the first access server to receive a packet and establish a link, this access server creates a virtual interface and becomes the bundle master. The bundle master takes ownership of the MLP session with the remote device. The Multichassis Multilink PPP (MMP) protocol forwards the second link from device #2 to the bundle master, which in turn bundles the two B channels together and provides 128 kbps to the end user. Layer 2 Forwarding (L2F) is the mechanism that device #2 uses to forward all packet fragments received from the terminal adapter to device #1. In this way, all packets and calls virtually appear to terminate at device #1.

Figure 5 A Stack Group of Access Servers Using MMP Without an Offload Processor

Setting Up an Offload Server

Because MMP is a processor-intensive application, you might need to offload the processing or segmentation and reassembly from the Cisco access servers to a router with a more powerful CPU, such as the Cisco 4700-M or Cisco 7206. We recommend that you include an offload server for dial-in solutions that support more than 50 percent ISDN calls or more than 10 multilink sessions per Cisco access server. (See Figure 6.)

Figure 6 A Stack Group of Access Servers Using MMP with an Offload Processor

Using the Stack Group Bidding Protocol

The Stack Group Bidding Protocol (SGBP) is a critical component used in multichassis multilink sessions. SGBP unites each Cisco access server in a virtual stack, which enables the access servers to become virtually tied together. Each independent stack member communicates with the other members and determines which devices’ CPU should be in charge of running the multilink session and packet reassembly—the duty of the bundle master. The goal of SGBP is to find a common place to forward the links and ensure that this destination has enough CPU power to perform the segmentation and packet reassembly. (See Figure 6.)

When SGBP in configured on each Cisco access server, each access server sends out a query to each stack group member stating, for example, “I have a call coming in from walt@options.com. What is your bid for this user?” Each access server then consults the following default bidding criteria and answers the query accordingly:

• Do I have an existing call or link for the user walt@options.com? If I do, then bid very high to get this second link in to me.
• If I do not have an existing call for walt@options.com, then bid a value that is proportional to how much CPU power I have available.
• How busy am I supporting other users?

Note An offload server will always serve as the bundle master by bidding a higher value than the other devices.

Using L2F

Note Effective with Cisco Release 12.4(11)T, the L2F protocol was removed in Cisco IOS software.

L2F is a critical component used in multichassis multilink sessions. If an access server is not in charge of a multilink session, the access server encapsulates the fragmented PPP frames and forwards them to the bundle master using L2F. The master device receives the calls, not through the dial port (such as a dual T1/PRI card), but through the LAN or Ethernet port. L2F simply tunnels packet fragments to the device that owns the multilink session for the call. If you include an offload server in your dial-in scenario, it creates all the virtual interfaces, owns all the multilink sessions, and reassembles all the fragmented packets received by L2F via the other stackgroup members. (Refer to Figure 6.)

Stack Group of Access Servers Using MMP with an Offload Processor Examples

The following sections provide examples for the devices shown in Figure 6:

• Cisco Access Server #1
• Cisco Access Server #2
• Cisco Access Server #3
• Cisco 7206 as Offload Server
• RADIUS Remote Security Examples

Note Be sure to include your own IP addresses, host names, and security passwords where appropriate.

Cisco Access Server #1

The following configuration runs on the Cisco access server labeled #1 in Figure 6:

Cisco Access Server #2

The following configuration runs on the Cisco access server labeled #2 shown in Figure 6:

Cisco Access Server #3

The following configuration runs on the Cisco access server labeled #3 in Figure 6:

Cisco 7206 as Offload Server

The following configuration runs on the Cisco 7206 router shown in Figure 6:

Note Any Cisco router that has a powerful CPU can be used as an offload server, such as a Cisco 4500-M, 4700-M, or 3640. However, the router must be configured to handle the necessary processing overhead demanded by each stack member.

RADIUS Remote Security Examples

The RADIUS examples in the following sections use the Internet Engineering Task Force (IETF) syntax for the attributes:

• User Setup for PPP
• User Setup for PPP and Static IP Address
• Enabling Router Dial-In
• User Setup for SLIP
• User Setup for SLIP and Static IP Address
• Using Telnet to connect to a UNIX Host
• Automatic rlogin to UNIX Host

Depending on how the dictionary is set up, the syntax for these configurations might differ between versions of RADIUS daemons.

Note You must have the async dynamic address command enabled on the network access server if you use Framed-IP-Address to statically assign IP addresses.

User Setup for PPP

The following example shows a user setup for PPP. The user’s IP address comes from the configured default IP address that is set up on the interface (which could be a specific default IP address, a pointer to a local pool of addresses, or a pointer to a Dynamic Host Configuration Protocol (DHCP) server). The special address that signals the default address is 255.255.255.254.

User Setup for PPP and Static IP Address

The following example shows a user setup for PPP and a static IP address that stays with the user across all connections. Make sure that your router is set up to support this configuration, especially for large or multiple POPs.

Enabling Router Dial-In

The following example supports a router dialing in, which requires that a static IP address and a remote Ethernet interface be added to the network access server’s routing table. The router’s WAN port is assigned the address 1.1.1.2. The remote Ethernet interface is 2.1.1.0 with a class C mask. Be sure your routing table can support this requirement. You might need to redistribute the static route with a dynamic routing protocol.

User Setup for SLIP

The following example shows a user setup for SLIP. Remote users are assigned to the default address on the interface.

User Setup for SLIP and Static IP Address

The following example shows a user setup for SLIP and a static IP address that stays with the user across all connections. Make sure that your routing is set up to support this configuration, especially for large or multiple POPs.

Using Telnet to connect to a UNIX Host

The following example automatically uses Telnet to connect the user to a UNIX host. This configuration is useful for registering new users, providing basic UNIX shell services, or providing a guest account.

Automatic rlogin to UNIX Host

The following example automatically uses rlogin to connect the user to a UNIX host:

If you want to prevent a second password prompt from being brought up, you must have the following two commands enabled on the router or access server:

• rlogin trusted-remoteuser-source local
• rlogin trusted-localuser-source radius

Overview

Many cities throughout the world have large installed bases of PCs that interface with older modems, PADs, and X.25 networks. These remote PCs or terminals dial in to PADs and make X.25 PAD calls or terminal connections to mainframe computers or other devices, which run the X.25 protocol. Unfortunately, the user interface is only a regular text-based screen in character mode (as opposed to packet mode). Therefore, many ISPs and telcos that have large investments in X.25 networks are upgrading their outdated equipment and creating separate networks for PPP connections. Because this upgrade process takes substantial time and money to complete, using a Cisco router to allow PPP connections over an X.25 network is a good interim solution for a dead-end dial case.

Remote PC Browsing Network Topology

Figure 7 shows a remote PC browsing the Internet through an X.25 PAD call and a Cisco 4500 router. This X.25 network is owned by an ISP or telco that is heavily invested in X.25 equipment, that is currently upgrading its outdated equipment, and that is creating separate networks for PPP connections. In this topology, the Cisco 4500 router performs protocol translation between the protocols X.25 and PPP. The router is configured to accept an incoming X.25 PAD call, run and unpack PPP packets over the call, and enable the remote PC to function as if it were on the IP network.

Figure 7 Remote PC Browsing the Internet Through an X.25 PAD Call and a Cisco 4500 Router

For more information about configuring protocol translation, see the chapter “Configuring Protocol Translation and Virtual Asynchronous Devices” in the Cisco IOS Terminal Services Configuration Guide.

Protocol Translation Configuration Example

In the following example, PAD callers that dial 4085551234 receive a router prompt. PAD callers that dial 4085555123401 start PPP and pick up an address from the IP pool called dialin_pool. These addresses are “borrowed” from the Ethernet interface on the Cisco 4500 router. Additionally, a loopback interface network can be created and the X.25 addresses can be set. However, a routing protocol must be run to advertise the loopback interface network if this method is used.

Note Be sure to include your own IP addresses, host names, and security passwords where appropriate in the following examples.

Cisco and the Cisco logo are trademarks or registered trademarks of Cisco and/or its affiliates in the U.S. and other countries. To view a list of Cisco trademarks, go to this URL: www.cisco.com/go/trademarks. Third-party trademarks mentioned are the property of their respective owners. The use of the word partner does not imply a partnership relationship between Cisco and any other company. (1110R)