Distributed Virtual Routing (DvR) is a technology for router redundancy in a fabric deployment where IP subnets are stretched across multiple switches. DvR provides Default Gateway Redundancy and optimizes traffic flows to avoid traffic tromboning due to inefficient routing, thereby increasing the total routing throughput.
DvR can be deployed in Campus environments for stretching IP subnets between multiple aggregation layer switches and also simplifies data center deployments by introducing a Controller-Leaf architecture. In this architecture, Layer 3 configuration is required only on the Controller nodes, whereas the Leaf nodes require only Layer 2 configuration. All Layer 3 configuration is automatically distributed to the Leaf nodes by the Controller nodes.
For typical Campus DvR deployments, configure aggregation layer switches as DvR Controllers. Wiring closet access switches are then typically dual-homed to a pair of DvR Controllers.
IP subnets, which stretch between aggregation layer switches and multiple wiring closets, enable seamless IP roaming for wireless users while at the same time ensure optimal traffic forwarding. To optimize automation, Fabric Attach is typically deployed between wiring closet and aggregation switches. In this construct, there would likely be no DvR Leaf configured.
In fabric deployments, DvR replaces VRRP (with VRRP-BackupMaster or RSMLT). The operator can chose for each I-SID/IP subnet what router redundancy method to use.
To migrate to a DvR-enabled I-SID/IP subnet, all member fabric switches of this I-SID must be either DvR Controllers or DvR Leafs. You can connect non fabric switches to DvR Leafs and DvR Controllers with manual configuration or Fabric Attach configuration. Until all fabric switches that are members of the I-SID/IP subnet are DvR-enabled, use VRRP or RSMLT as the router redundancy protocol.