For the purpose of high availability, critical locations of company A – a customer of VPN service provider – is connected to two different carriers. However, this connectivity requires an important design consideration.
In the network design shown above, customer AS 64512 is connected to two different providers: AS100 and AS200. Since this site is critical to the business of company A, it decides to use two routers and two links to connect to the service providers.
During the steady state of the network, CE2 uses AS200 to reach CE5, which is single homed and connected to only AS200. In fact, CE2 can only reach CE5 through AS200. If connectivity fails between CE2 and AS200 (as depicted in the Figure-2), CE5 cannot be reachable anymore even though CE5 can still be connected to AS200.
In order to connect to single homed sites such as CE5, transit site connected to both service providers are used.
In Figure-3, transit site (AS 64545) is connected to both service providers. In case CE2 connection fails, AS 64512 can still reach single homed CE5 through the transit site.
There are two important design considerations for the transit site: One is that bandwidth capacity should be arranged accurately so that in case of failure, there shouldn’t be bottleneck; the other is that transit site should be chosen based on the geography of the entire network. Doing so will reduce the latency between the single homed and the dual homed sites in case of failure.