BGP OSPF Traffic Engineering

[MEDIAOS]

See attached for diagram:

• R1 advertises a router to R2 via the MPLS using BGP – On R2, I am going to redistribute the route into OSPF. From R2 to R1 the BGP would be the preferred path – How do I force R2 to use the OSPF link as primary route for traffic to R1?
• R1 advertises a route to R2 via the MPLS using BGP – On R2, I am going to redistribute the route into OSPF --- On R2, how do I prevent OSPF from advertising R1 route back to R1.
• R2 advertises a /24 (10.1.0.0/24) subnet to the MPLS network – R3 will advertise this route to R1. How can I tell R3 to advertise a /16 (10.1.0.0/16 le 32) to R1 (not the specific route the /24) keep in mind R3 does not originate the route?

[MHM Cisco World]

..

[MEDIAOS]

Hello MHM, I cannot use back door because I would like R2 to use the MPLS to route traffic to R2. (R2 and R3 are in the same geographical location)

See attached diagram.

[MHM Cisco World]

I could not open ZIP in may MacBook, can you share it as photo?

[MEDIAOS]

[MHM Cisco World]

...

[MEDIAOS]

Yes, i would like R2 to use the direct link to R1 (Which is running OSPF). Because of the lower AD, I think R2 would prefer the MPLS link.

[MHM Cisco World]

So R2 is CE and have two path for prefix in R1 
one via eBGP "between CE-R2 and MPLS Core" which have 20 and other is direct connect OSPF which have 90 ??
using backdoor, 
see this link below both case using backdoor and sham-link.

https://networkinferno.net/ccie-study-ospf-sham-link

[MHM Cisco World]

R2 advertises a /24 (10.1.0.0/24) subnet to the MPLS network – R3 will advertise this route to R1. How can I tell R3 to advertise a /16 (10.1.0.0/16 le 32) to R1 (not the specific route the /24) keep in mind R3 does not originate the route?

I think you config redundancy by this way
ip route 10.1.0.0/16 null0 <- in R3 

router bgp 
net 10.1.0.0 mask 255.255.0.0

this make R3 advertise /16 to R1 and still R1 receive /24 from R2 and prefer it "longest match of routing table"
when this /24 prefix disappear then still R1 have path via R3 /16 

[David Ruess]

1.) You could lower the AD of OSPF to choose that first

2.) You could add a route tag to the redistributed routes on R2 and filter them out when they try to leave the OSPF interface back to R1.

3.) YOU could do a BGP summary route on exit interface of R3 with extras:

aggregate-address 10.1.0.0 255.255.255.0 summary-only advertise-map <route-map-name> (route map references prefix list of le 16)

[paul driver]

Hello
A possible alternative if I understand your OP could be to use a condition within bgp, so if certain prefixes are seen within the bgp rib of R3 then withdraw an advertised prefix to R1

Example
R2 advertise 10.1.0.1/32
R3 to advertise 10.1.0.0/28 to R1 if 10.1.0.1/32 is not seen in its bgp rib table sent by R2

R3
access-list 10 permit host 10.1.0.1
access-list 20 permit host 10.1.0.0 0.0.0.15

route-map missing-routes
match ip address 10

route-map advertise-route
match ip address 20

router bgp xx
neighbor (r1)  advertise-map advertise-route non-exist-map missing-routes