Redistribution of Routes to OSPF or BGP

For non-SPB routers to benefit from the host accessibility information learned within a DvR domain, DvR supports the redistribution of host routes into OSPF or BGP. Redistribution of these host routes is only by the DvR Controllers and only for the intra-domain host routes within the DvR enabled subnets.

A DvR Controller can redistribute host routes for all hosts from a DvR domain into OSPF or BGP. You can also apply route policies on the Controller to select the routes to be redistributed. The Controller supports redistribution of routes from the Global Router or within a VRF instance. You can also configure the metric of the route before redistribution.

The following example demonstrates the benefit of redistribution of routes to BGP.

Consider a 10.1.0.0/16 network with a stretched Layer 2 VSN spanning two data centers. On the campus side of the network, BGP peering is configured between a non-Extreme router and one or more routers in the data center. BGP advertises the network route 10.1.0.0/16 to the campus BGP routers. Depending on which edge router the traffic is delivered to, it is possible that traffic from a host on the campus traverses the WAN a second time to reach the server that is physically connected to one segment of the data center, as shown in the following figure.

Click to expand in new window
Inefficient traffic flow

Redistribution of the host routes from the DvR Controller to BGP solves this problem.

The following figure shows two DvR domains (show in green and orange) configured at each data center. Each campus edge router establishes a BGP peering session with one or more Controllers in each data center (DvR domain). This enables BGP to advertise more specific routes to the campus BGP router so that the optimal routing path is always taken. So, there is no need for traffic to traverse the WAN multiple times. Also, in the case of server movement within or between data centers, the updated DvR host routes are propagated to BGP, thus ensuring that traffic flowing into the data center continues along the most optimal path.

For example, in the following figure, only the Controller attached to the Leaf node where the 10.1.0.111 server exists, advertises its accessibility over the 10.1.0.111/32 route. Similarly, the DvR Controller associated with the Leaf node connected to the 10.1.0.222 server advertises the 10.1.0.222/32 host route.

Click to expand in new window
Traffic flow optimized with route redistribution

Controllers in each data center learn all host routes through the DvR backbone, but since those routes belong to different DvR domains, they are not all eligible for redistribution to OSPF or BGP.