 |
Note: Remember to consider parameters such as link bandwidth, QoS, and line module scaling limitations when you plan the deployment of the ICR cluster. |
An ICR cluster is a group of routers participating in ICR deployment. When planning an ICR cluster you must ensure that you have adequate resources in the event of a worst-case failure scenario such as a multiple hardware or multiple router failure. For instance, most networks can handle the failure of a single router. However, they may not be able to handle multiple router failures. ICR enables you to choose the degree of redundancy in your ICR cluster. Depending on the type of network that you have, you can design a 1:1 (minimum) or 1:N (maximum) degree of redundancy in the ICR cluster.
|
Note: Remember to consider parameters such as link bandwidth, QoS, and line module scaling limitations when you plan the deployment of the ICR cluster. |
Consider a 4–node ICR cluster that consists of four ERX1440 routers, as shown in Figure 8. Each of the four routers is capable of supporting 48,000 PPP/PPPoE subscribers. The degree of redundancy that you can achieve in this cluster is 1:1. For every subscriber, you have a backup destination within the cluster. If one router fails, subscriber load is equally distributed to the other three routers. Thus, no single router serves as a dedicated backup. Instead, each router is loaded with 75 percent of its capacity and the remaining 25 percent is kept unused to accommodate subscribers from the failing router. Failure of any one router causes all routers in the cluster to become fully loaded with no spare capacity to accommodate further failures. This is the minimum degree of redundancy in a 4–node ICR cluster.
Figure 8 illustrates an example of 1:1 redundancy.
Figure 8: Sample 1:1 Subscriber Redundancy in a 4–Node ICR Cluster
You can increase the redundancy in a cluster by configuring more than one backup destination for each subscriber. You can achieve a fully redundant system in which a single router can support all the subscribers when all the other routers fail. To achieve this, you may have to compromise the operating efficiency of the deployed hardware.
Consider the same 4–node ICR cluster but where each router is loaded with 25 percent of its actual capacity as shown in Figure 9. If three routers of the four node cluster fail, the single router has enough spare capability to accommodate the entire subscriber load. This is the maximum degree of redundancy in a 4–node ICR cluster.
Figure 9 illustrates an example of 1:3 redundancy.
Figure 9: Sample 1:3 Subscriber Redundancy in a 4–Node ICR Cluster
 |
Copyright © 2009, Juniper Networks, Inc. All rights reserved. Trademark Notice. | |