This chapter is from the book
This chapter is from the book
This chapter is from the book
2.3 Load Sharing
Note that in Figure 2-4, R2 router is completely idle during its backup periods. Its whole purpose in the network consists of being a backup for R1. R2 is a purely redundant device. In certain circumstances this may not be the best arrangement. Leaving a network device idle, in particular in the case of a robust master, may be considered underutilization or waste of a valuable resource. In such circumstances the network deployer may decide to assign R2 as a default router to some hosts on the LAN—say, H3 and H4. With this configuration traffic from H1, H2 is forwarded to R1 and traffic from H3 and H4 to R2.
){kind=link}
The obvious advantage of this configuration is the establishment of a load-sharing scheme. With this configuration, the traffic originating from N1 network is not sent exclusively to one of the routers but is shared between R1 and R2. The traffic coming from H1 and H2 is handled by R1, and H3-H4 traffic is forwarded to R2.
But it must be realized that the virtual router configuration we have been discussing so far would not be of help to protect R2 in its default first hop router role, since being a member of a virtual router does not imply protection for the interfaces of a VRRP router. The protection needs to be explicitly set up.
To create this setup, we need to define two virtual routers, V1 and V2, and we need to define the opposite roles to our routers in V1 and V2. Figure 2-5 illustrates this setup.
)
FIGURE
2-5. Load sharing between VRRP routers
In this setup R1 is defined as the master of the V1 and R2 as the backup. In V2, R2 is the master and R1 acts as the backup. With this configuration, we establish a load-sharing arrangement between R1 and R2; moreover, we create a mutual protection setup by having two routers acting as backups for each other. Note, however, that in the case of the failure of one of the routers, one can observe some degradation in the network service, unless the network is engineered accordingly.
Although many network deployers expressed to us their liking for the load-sharing aspect of this configuration and were agreeable to suffering some level of loss in the quality of service, the service degradation was not acceptable to some other customers, and these others had rather a strong preference for a one-way protection scheme, even at the expense of leaving a resource completely idle during the backup periods. Our experience votes for load sharing. Often, backup equipment does not work when needed just because it is not kept in a fully working state. Any equipment that is not performing its primary function may be subject to administrative neglect and fail to get proper maintenance or required software/hardware upgrades. This is a major risk, defeating the purpose of the whole scheme.
Another advantage of using both routers at the same time is that this approach ensures that the network manager knows when one fails. It is much easier to detect the failure of an active device than of equipment that is in a passive monitoring mode.