A TFTP server, DHCP server, and time-of-day (TOD) server are required to support two-way data cable modems on an HFC network. A cable modem will not boot if these servers are not available. The log server and security servers are not required to configure and operate a cable modem. If the log server or security servers are not present, a cable modem will generate warning messages, but it will continue to boot and function properly.

Figure 1. Servers Required on a Two-Way HFC Network

In this provisioning model, TOD and TFTP servers are standard Internet implementations of the RFC 868 and RFC 1350 specifications. Most computers running a UNIX-based operating system supply TOD and TFTP servers as a standard software feature. Typically, the TOD server is embedded in the UNIX inetd and it requires no additional configuration. The TFTP server is usually disabled in the standard software but can be enabled by the user. Microsoft NT server software includes a TFTP server that can be enabled with the services control panel. Microsoft NT does not include a TOD server. A public domain version of the TOD server for Microsoft NT can be downloaded from several sites.

The DHCP and Domain Name System (DNS) server shown in Figure above must be the DHCP/DNS server available in Cisco Network Registrar version 2.0 or later. CNR is the only DHCP server that implements policy-based assignment of IP addresses. The headend must be a Cisco cBR-8 converged broadband router. The remote access server is only required on HFC networks that are limited to one-way (downstream only) communication. In a one-way HFC network, upstream data from a PC through the headend to the Internet is carried over a dialup connection. This dialup connection for upstream data is referred to as telco return. For simplification, the model will not include a log or security server. Cable modems can be set up to use the logging and security servers by including the appropriate DHCP options in the cable modem policy as described in the Cisco Network Registrar User Manual.

CNR is a dynamic IP address management system, running on Windows or Solaris, that uses the Dynamic Host Configuration Protocol (DHCP) to assign IP addresses to cable interfaces, PCs, and other devices on the broadband network. The CNR tool includes script extensions that allow a cable system administrator to define and view individual DHCP options, define the identity or type of device on the network, and assign the device to a predefined class or group.

Using the CNR tool, a cable system administrator can specify policies to provide:

• Integrated DHCP and Domain Name Server (DNS) services
• Time of Day (ToD) and Trivial File Transfer Protocol (TFTP) server based on the size of the network
• DHCP safe failover and dynamic DNS updates

Note	This is available only in CNR 3.0 or higher.

Using the CNR tool and the extension scripts identified in the Overview of Scripts section, a cable system administrator can specify scopes, policies, and options for the network and each cable interface based on the services and configuration to support at each subscriber site.

Note	Scopes refer to the administrative grouping of TCP/IP addresses; all IP addresses within a scope should be on the same subnet.

The cable system administrator defines system default policies for all standard options and uses scope-specific policies for options related to particular subnets, such as cable interfaces. This allows DHCP to send the information with the IP address.

Seven entry points exist for scripts:

• post-packet-decode
• pre-client-lookup
• post-client-lookup—Examines and takes action on results of the client-class process, places data items in the environment dictionary to use at the pre-packet-encode extension point, includes DHCP relay option
• check-lease-acceptable
• pre-packet-encode
• post-sent-packet
• pre-dns-add-forward

Cisco Network Registrar (CNR) is a dynamic IP address management system that uses the Dynamic Host Configuration Protocol (DHCP) and assigns IP addresses to PCs and other devices on a network based on a predefined set of policies, such as class of service. CNR assigns available IP addresses from address pools based on the identity or type of the requesting device and the policies in effect. For example, CNR can distinguish between registered devices, unregistered devices, and registered devices that have been assigned to a particular class of service.

CNR also provides extensions that can be customized (via programming or a script) so that you can view individual DHCP options, determine the identity or type of a device based on the content of the options, and assign a device to a predefined class or group. Using these extensions, you can determine the difference between PCs and cable modems and assign them IP addresses from different address pools.

In typical data-over-cable environments, service providers are interested in simplifying provisioning to limit the amount of information that must be collected about subscribers’ customer premise equipment (CPEs). To support current provisioning models, a field technician must be sent to a subscriber’s home or business to install and setup a cable modem. During this site visit, the technician might register the serial number and MAC address of the cable modem in the customer account database. Because a field technician must go to a subscriber’s site to replace a cable modem, you can easily track modem information.

Manually registering and tracking information about a cable subscriber’s PC is more difficult. A subscriber might purchase a new PC or exchange the network interface card (NIC) without notifying you of the change. Automatic provisioning with CNR reduces the amount of customer service involvement needed to track customer equipment. To use the provisioning model described in this document, you must still track serial numbers and MAC addresses for cable modems, but you do not need to track information about the PC or NIC cards installed at a subscriber site.

The remainder of this document describes how to configure CNR to support this model. The following sections describe the equipment and servers required for the cable headend, provide an overview of the interaction between DOCSIS-compatible cable modems and the Cisco universal broadband routers, and provide a guide on how to configure CNR to support this provisioning model.

Cisco converged broadband routers and cable modems are based on the Data Over Cable Service Interface Specification (DOCSIS) standards. These standards were created by a consortium of cable service providers called Multimedia Cable Network Systems, Ltd. (MCNS) to that cable headend and cable modem equipment produced by different vendors will interoperate. The key DOCSIS standards provide the basis for a cable modem to communicate with any headend equipment and headend equipment to communicate with any cable modem.

Cable modems are assigned to operate on specific cable channels so activity can be balanced across several channels. Each Cisco cBR-8 router installed at the headend serves a specific channel. Part of network planning is to decide which channel each cable modem can use.

A cable modem cannot connect to the network until the following events occur:

• The cable modem initializes and ranges through available frequencies until it finds the first frequency that it can use to communicate to the headend. The cable modem might be another vendor’s DOCSIS-compatible device and the headend might have a Cisco cBR-8 router installed. At this point on the initial connection, the cable modem cannot determine if it is communicating on the correct channel.
• The cable modem goes through the DHCP server process and receives a configuration file from the server.
• One of the parameters in the configuration file tells the cable modem which channel it can use.
• If the assigned channel is not available on the Cisco cBR-8 router to which the cable modem is currently connected, it resets itself and comes up on the assigned channel.
• During this second DHCP process, the modem will be connected to the correct CMTS. This time, the configuration file will be loaded. For a DOCSIS-compatible cable modem to access the network, it might go through the DHCP server two times on two different networks; therefore, one-lease-per-client IP addressing is critical.

DHCP options and packet fields are required to enable cable modems to boot and operate properly. Table below lists the required DHCP options and fields.

You can use the following information to set up Cisco Network Registrar in a trial configuration. The configuration describes DHCP-related setup only; it does not cover setting up DNS or configuring dynamic DNS (DDNS). You should be familiar with important CNR concepts including scopes, primary and secondary scopes, scope selection tags, client classes, and CNR policies. See the Using Network Registrar publication for detailed information on these concepts.

In the trial configuration, you can configure CNR to perform the following operations:

• Receive DHCP requests from a cable modem and a PC on an HFC network via a port supporting multiple network numbers. The Cisco cBR-8 router at the headend must be configured as a forwarder (iphelper is configured).
• Serve IP addresses on two networks; a net-10 network (non-Internet routable) and a net-24 network (Internet routable).
• Tell the difference between a cable modem and a PC based on the MAC address of the device and provide net-24 addresses to the PC and net-10 addresses to the cable modem.
• Refuse to serve IP addresses to MAC addresses that it does not recognize.

To perform these options, you must implement the following CNR configuration items:

• Create two scope selection tags; one for PCs, one for cable modems.
• Create two client-classes; one for PCs , one for cable modems.
• Create a lease policy appropriate for the cable modem devices.
• Create a lease policy appropriate for the PC devices.
• Create a scope containing Class A net-24 (routable) addresses.
• Create a scope containing Class A net-10 (nonroutable) addresses.
• Identify the scope containing the net-24 addresses as the primary scope and configure the other scope containing the net-10 addresses as secondary to the net-24 scope.

Note	The Cisco cBR-8 router upstream ports must be configured with the primary network address on the net-24 network; such as 24.1.1.1.

• Assign the policies to the appropriate scope.
• Add the MAC address of the cable modem and the PC to the client-entry list.
• Associate the PC tag with the scope containing routable addresses.
• Associate the cable modem tag with the scope containing nonroutable addresses.
• Associate the cable modem tag with the cable modem client-class.
• Associate the PC tag with the PC client-class.
• Assign the PC MAC to the PC class.
• Assign the cable modem MAC to the cable modem class.
• Enable client-class processing.

Figure below shows the trial CNR configuration in an HFC network.

Figure 2. Trial Configuration in an HFC Network

These configuration items and their associations can be created using either the CNR management graphical user interface (GUI) or command-line interface (CLI). The following sample script configures DHCP for a sample server:

File: cabledemo.rc
Command line: nrcmd -C <cluster> -N <user name> -P <password> -b < cabledemo.rc
---------------------------------------------------------------------------------------
scope-selection-tag tag-CM create
scope-selection-tag tag-PC create
client-class create class-CM 
client-class class-CM set selection-criteria=tag-CM
client-class create class-PC 
client-class class-PC set selection-criteria=tag-PC
policy cmts-cisco create
policy cmts-cisco setleasetime 1800
policy cmts-cisco setoption domain-name-servers 192.168.10.2
policy cmts-cisco setoption routers 10.1.1.1
policy cmts-cisco setoption time-offset  604800
policy cmts-cisco setoption time-servers 192.168.10.20
policy cmts-cisco set packet-siaddr=192.168.10.2
policy cmts-cisco setoption log-servers 192.168.10.2
policy cmts-cisco setoption mcns-security-server 192.168.10.2
policy cmts-cisco set packet-file-name=golden.cfg
policy cmts-cisco set dhcp-reply-options=packet-file-name,packet-siaddr,mcns-security-server
policy pPC create
policy pPC set server-lease-time 1800
policy pPC setleasetime 1800
policy pPC setoption domain-name-servers 192.168.10.2
policy pPC setoption routers 24.1.1.1
scope S24.1.1.0 create 24.1.1.0 255.255.255.0
scope S24.1.1.0 addrange 24.1.1.5 24.1.1.254
scope S24.1.1.0 set policy=pPC
scope S24.1.1.0 set selection-tags=tag-PC
scope S10.1.1.0 create 10.1.1.0 255.255.255.0
scope S10.1.1.0 addrange 10.1.1.5 10.1.1.254
scope S10.1.1.0 set policy=cmts-cisco
scope S10.1.1.0 set selection-tags=tag-CM
scope S10.1.1.0 set primary-scope=S24.1.1.0
client 01:02:03:04:05:06 create client-class-name=class-PC
client ab:cd:ef:01:02:03 create client-class-name=class-CM
client default create action=exclude
dhcp enable client-class
dhcp enable one-lease-per-client
save
dhcp reload

In addition to the DHCP server setup, you might want to enable packet-tracing. When packet-tracing is enabled, the server parses both requests and replies, and then adds them to the logs. If you do enable tracing, performance will be adversely affected, and the logs will roll over quickly.

Use the following nrcmd command to set packet tracing.

DHCP set log-settings=incoming-packet-detail,outgoing-packet-detail

This section lists the scripts applicable to cable interface configuration.

• Windows NT
• Solaris

\program files\network registrar\extensions\dhcp\scripts\tcl

/opt/nwreg2/extensions/dhcp/scripts/tcl