High availability for data center interconnect WAN links is significantly more complex than it is for a LAN environment due to factors such as distance, cost, non-ownership of equipment, and connectivity. The HA model for WAN connectivity varies widely depending on the type of WAN being employed. In general, a typical WAN HA solution can be classified as either including redundant components, redundant links, or both.

Redundant components means having multiple devices, such as routers or multiplexers, at each site. This provides redundancy in the event of component failure and allows for non-disruptive maintenance and upgrades. Redundant links refers to having multiple connections between data centers. This can be achieved through multiple dedicated media, such as multiple strands of dark fibre, or through a combination of fiber and a shared medium, such as MPLS. Provide a combination of both methodologies to have a fully redundant data path for the highest level of availability.

Some service providers might also provide the option of having a standby connection. This option often has a lower total cost of ownership because you are charged by the amount of data transmitted. This might be a cost-effective way of achieving a redundant link without having to pay for unrequired bandwidth.

Before making the appropriate design decision about WAN availability, understand the available options and what needs to be protected against. For example, deploying redundant links or components does not provide any benefit if the single ISP has a large-scale outage.