Guidelines for Deploying an ICR Partition in Your Network

[Contents] [Prev] [Next] [Index] [Report an Error]

To support 1: N redundancy and to allow your network to scale, you must deploy an ICR partition in your network. An ICR partition is a logical group of subscribers each of which you can manage using a unique VRRP instance. You can configure multiple partitions on a single physical interface to allow your network to distribute subscriber load to multiple backup destinations.

In a stateless model, when a partition fails, all subscribers belonging to the failing partition eventually time out and must log in again to a new active destination based on the state of the VRRP instance. A partition in the backup state does not accept subscriber login requests. Also, in a stateless model, the router sends early termination requests to clients so that clients do not wait for timeout conditions to occur in order to send requests to log in again. In this way, a stateless model is best suited for PPP/PPPoE subscribers.

Note: To avoid routing issues while configuring stateless ICR for PPPoE subscribers, you must configure non-overlapping IP addresses for the clients. This indicates that the client is always assigned a different IP address after failover.

Hardware Requirements for ICR

Before you deploy stateless ICR for PPP/PPPoE subscribers, you must plan and obtain hardware resources to fulfill QoS and bandwidth requirements. You can identify the hardware required based on the type of ICR cluster planned. For instance, if you plan to create a simple ICR cluster that provides 1:1 redundancy, the minimum hardware capability you must plan for is 2. 1:1 redundancy implies that for every subscriber, you have a backup destination within the cluster. Each physical interface configured as part of the ICR cluster must have one backup destination interface on the other router. Also, the router on which you configure the backup partition must have enough bandwidth to accommodate new subscribers after failover.

Network Requirements for ICR

After you plan for and obtain hardware for ICR, you must create the network topology and connections in order to set up a broadcast network between the master and backup interfaces. You can use layer 2 switches and configure them to provide selective broadcast connectivity using VLAN tags. The new broadcast network must allow multicast VRRP traffic between participating ICR interfaces.

Router Configurations for ICR

After setting up the broadcast network, you can configure AAA, QoS, and the interface on the router. You must also configure the uplink interfaces separately on each router that is part of the ICR cluster. Uplink interfaces do not have backups; they behave as if they have been configured on two independent routers.