In a DvR domain, DvR enabled Controller(s) handle the learning and distribution of Layer 3 configuration and route data to the DvR enabled Leaf nodes. The Leaf nodes in turn, use this data to automatically create distributed Layer 3 datapaths on themselves. In this way, Layer 3 configuration and learning remains only with the Controller(s) and there are distributed Layer 3 datapaths at the edges of the fabric network. This allows for destination lookups at the edge to happen quickly, and traffic is sent directly to their destinations without multiple lookups.
An important benefit of DvR is that only minimal configuration is required on the Leaf node. Based on the Layer 2 VSN that the Leaf node is a part of, all Layer 3 configuration information (IPv4 Unicast and Multicast configuration) is pushed from the Controllers in the domain. Thus the leaf nodes, although basically Layer 2 configured switches, become fully layer 3 capable devices.
