Access Redundancy

The access layer presents many of the same challenges and issues as the distribution layer, and it also shares some of the same strategies for resolving these drawbacks. Dual homing access layer devices are the most common way of providing redundancy to remote locations, but it's also possible to interconnect access layer devices to provide redundancy.

In Figure 3-10, Router G and Router H are access layer routers that are dual-homed with the backup circuit connected to different branches of the distribution layer. If these redundant links are actually constantly up and carrying traffic, the number of paths between 10.2.1.0/24 and 10.1.1.0/24 is excessive:

Figure 3-10 Access Layer Redundancy—Dual Homing through Different Distribution Branches

Figure 3-10 Access Layer Redundancy—Dual Homing through Different Distribution Branches

Access Layer Router

• Router H to Router F to Router B to Router A to Router C to Router G

• Router H to Router F to Router B to Router E to Router G

• Router H to Router D to Router A to Router B to Router E to Router G

• Router H to Router D to Router A to Router C to Router G

With each addition of a dual-homed access layer router, things get worse. This plethora of paths causes major problems in the core; the size of the routing table in the core will mushroom. This is the general rule: If the redundant link crosses the boundary of a distribution layer branch, it should not be advertised as a normal path.

Another option to provide access layer redundancy (and another illustration of the general rule above) is to provide links between the access layer routers themselves. In Figure 3-11, this saves one link, and it also reduces the number of paths between 10.1.1.0/24 and 10.2.1.0/24 down to two. If access layer redundancy is provided using links between access devices, it's important to provide enough bandwidth to handle the traffic from both remote sites toward the core.

Figure 3-11 Redundancy through Interconnected Access

Layer Devices

Figure 3-11 Redundancy through Interconnected Access

Either of these solutions would work well as long as the redundant route is not advertised until needed, so traffic won't normally flow across the redundant link. Dial-on-demand circuits work well for these types of applications.

It is possible to design load sharing and redundancy within the access layer, as Figure 3-12 illustrates. In this case, both links to Router G are connected to routers within the same distribution layer branch, as are both links to Router H.

Figure 3-12 Access Layer Redundancy through the Same

Distribution Layer Branch

Figure 3-12 Access Layer Redundancy through the Same

Distribution Layer Branch

It's still possible for packets traveling from Router C to Router D to pass through Router G, but this can be remedied with route filtering. Router G and Router H should only advertise the networks below them in the hierarchy. In Figure 3-12, this is 10.1.1.0/24 for Router G and 10.2.1.0/24 for Router H. If correct filtering is installed in Router G, Router C will not learn any paths through Router D by way of Router G.

One way to get around all of the problems associated with dual homing is to use dial backup. There are two sections at the end of this chapter, "Case Study: Dial Backup with a Single Router" and "Case Study: Dial Backup with Two Routers," that cover these options.

Was this article helpful?

0 0

Post a comment