Here's my situation. We have about 80 sites that are on MPLS. These sites backhaul through our main data center. We have around 150 SDWAN sites. In order for the SDWAN sites to reach the MPLS sites, they come back through the head end SDWAN router that is also in our data center. The SDWAN routing uses static routes, on our core DC switches. I have BGP setup, on our core DC switches, which peer with our DC MPLS router. Our DC BGP configuration advertises these static routes so that our MPLS sites can route to our SDWAN sites.
A few of our sites have a hybrid setup with a primary MPLS circuit, and a backup SDWAN circuit. At these sites, i'm not using a routing protocol. I'm simply using IP SLA with tracking and a default route with tracking, as well as a weighted floating static. Let's say that MPLS fails at this site. The tracking reachability goes down, so the static default is removed from the routing table, while the floating static is injected into the routing table. In our DC, we also have a floating static for the SDWAN backup for this site(with a 200). Again, when this site MPLS circuit/router go down, it removes the route from BGP, so that the floating static, in our DC, is injected into the routing table. Now when the MPLS side comes up again, the site tracking comes up, removing the floating static, and injecting the tracked default route. However, in our DC, the BGP route is never injected into the routing table. The route with the 200 stays as the primary route. The only way BGP updates with the BGP route, is if I manually remove the static 200. I'm assuming that I would have to setup local pref to change the behavior, but wanted to run this by you. Let me know what you think. Thanks!
