The corporate wide area network (WAN) is a funny thing. Even back in my early days as a network manager in the early 90s, there was talk of finding an alternative to the tried-and-true “hub and spoke” MPLS, frame relay, or other type of network. There’s no question that this type of network, although widely deployed, is inefficient, as it routes all traffic through a single choke point (the hub). Additionally, each branch location at the end of each spoke is at risk of being down if the WAN connection fails.
So why is this model of network design still so popular? Well, as inefficient as it is, it has worked OK with client/server-based applications, as most of these were in the corporate data center, or the hub. The inefficiency I spoke of was more relevant for Internet-based applications as that traffic had to come through a single point, traverse the WAN and then “trombone” back down the same connection.
Given the rise in cloud based applications, the need for WAN change has never been higher, and more and more companies are finally evolving the WAN. One of the more notable trends has been to connect branches with direct Internet access. The direct connection provides faster access to cloud applications and can also be used as an alternative connection to the primary network, creating a hybrid network.