The SAE uses the Common Open Policy Service (COPS) protocol as specified in the PacketCable Multimedia Specification PKT-SP-MM-I03-051221 to manage PCMM-compliant CMTS devices in a cable network environment. The SAE connects to the CMTS device by using a COPS over Transmission Control Protocol (TCP) connection. In cable environments, the SAE manages the connection to the CMTS device.
The CMTS device does not provide address requests or notify the SAE of new subscribers, subscriber IP addresses, or any other attributes. IP address detection and all other subscriber attributes are collected outside of the COPS connection to the CMTS device. The SAE uses COPS only to push policies to the CMTS device and to learn about the CMTS status and usage data.
Because the CMTS device does not have the concept of interfaces, the SRC software uses pseudointerfaces to model CMTS subscriber connections similar to subscriber connections for routers running JUNOS Software.
This section describes how the SAE is used in cable networks. It includes the following topics:
You can use two mechanisms to obtain subscriber address requests and other information and to set up a pseudointerface on the CMTS device. (You must choose one mechanism; you cannot mix them.):
With this method, you use the assigned IP subscriber login type along with the network interface collector (NIC) to map IP addresses to the SAE.
With this method, you use the event notification application programming interface (API). The API provides an interface to the IP address manager, and lets the IP address manager notify the SAE of events such as IP address assignments.
With the assigned IP subscriber method of logging in subscribers and creating sessions, the SRC software uses IP address pools along with NIC resolvers to provide mapping of IP addresses to SAEs. You configure the static address pools or dynamically discovered address pools in the virtual router configuration for a CMTS device. These pools are published in the NIC. The NIC maps subscriber IP addresses in requests received through the portal or Advanced Services Gateway to the SAE that currently manages that CMTS device.
This section describes login interactions for assigned IP subscribers. In the example shown in Figure 12, the subscriber activates a service through a portal. You could also have the subscriber activate a service through the Advanced Services Gateway.
Figure 12: Login Interactions with Assigned IP Subscribers

The sequence of events for logging in and creating sessions for assigned IP subscribers is:
Because the SAE is not notified when the subscriber logs out, the assigned IP idle timer begins when no service is active. The SAE removes the interface subscriber session when the timeout period ends.
With the event notification method of logging in subscribers and creating subscriber sessions, the subscriber logs in to the CMTS device and obtains an IP address through an address server, usually a DHCP server. The SAE receives notifications about the subscriber, such as the subscriber’s IP address, from an event notification application that is installed on the DHCP server.
To use this method of logging in subscribers, you can use the event notification API to create the application that notifies the SAE when events occur between the DHCP server and the CMTS device. You can also use Monitoring Agent, an application that was created with the event notification API, and that monitors DHCP or RADIUS messages for DHCP or RADIUS servers. See SRC PE Sample Applications Guide.
This section describes login interactions using event notifications.
Figure 13: Login Interactions with Event Notification Application

The sequence of events for logging in subscribers and creating sessions is:
The ipUp event should be sent with a timeout set to the DHCP lease time. The DHCP server sends an ipUp event for each Ack sent to the client. The SAE restarts the timeout each time it receives an ipUp event.
If the client explicitly releases the DHCP address (that is, it sends a DHCP release event), the DHCP server sends an ipDown event. If the client does not renew the address, the lease expires on the DHCP server and the timeout expires on the SAE.
For SAE redundancy in a cable network, you can have a community of two or more SAEs. SAEs in a community are given the role of either active SAE or passive SAE. The active SAE manages the connection to the CMTS device and keeps session data up to date within the community. Figure 14 shows a typical SAE community.
Figure 14: SAE Community

When an SAE opens a connection to the CMTS device, it negotiates with other SAEs to determine which SAE controls the CMTS device. The SAE community manager and members of the community select the active SAE.
A passive SAE needs to take over as active SAE in any of the following cases:
To aid in recovering from an SAE failover, the SAE stores subscriber and service session data. When the SAE manages a CMTS device, session data is stored locally in the SAE host’s file system. The SRC component that controls the storage of session data on the SAE is called the session store. The session store queues data and then writes the data to session store files on the SAE host’s disk. Once the data is written to disk, it can survive a server reboot.
For more information, see Fault Recovery.
To allow the SAE’s embedded policy server to communicate with a record-keeping server (RKS) in a PCMM environment, you need to use the PCMM record-keeping server plug-in. This plug-in is similar to the RADIUS accounting plug-ins, but it works with any RKS that is compliant with the PCMM specification. The RKS plug-in supports additional attributes: Application-Manager-ID, Request-Type, and Update-Reason. The plug-in sends all requests to the RKS as Acct-Status-Type=Interim-Update.