optimal Route reflector

[mautez_mah]

hi team
what is the difference between "optimal Route reflector" and "Route reflector"

[Giuseppe Larosa]

Hello @mautez_mah ,

can you report the link where you have found this "optimal route reflector" ?

In a service provider environment different design choices can be done in order to increase scalability:

a) to have dedicate route reflector servers for AF ipv4 unicast and dedicated route reflector servers for VPNv4

b) Additonal tips can be used like two level of route reflection for AF ipv4 unicast

c) for VPNv4 routes it is even possible to divide the load betweeen two dfferent subsets of RRS based on route target values

The leading principle for scalability is that non single device is required to store all the existing routes IPv4 or VPNv4.

I have read sometimes about the optimal position for a route reflector server and your link may refer to this specific part.

A route reflector can be a dedicated device with very powerful CPU and memory but not so strong in the forwarding plane or it can be a core device that is also acting as P node.

The use of dedicated devices for RRS role is typical.

Just to make an example an ASR 1000 with 16 GB of RAM can act as RRS.

Hope to help

Giuseppe

[mautez_mah]

https://www.cisco.com/c/en/us/support/docs/ip/border-gateway-protocol-bgp/212881-border-gateway-protocol-bgp-optimal-ro.html

[Giuseppe Larosa · Accepted solution]

Hello @mautez_mah ,

thanks for the link this is a special feature that is supported in IOS XR only.

>> The BGP Optimal Route Reflection solution allows the RR to send a specific best path to a specific BGP border router.

>> The solution allows for the best path calculation to be based on the BGP best path selection algorithm, which will lead to the RR picking the best path from the perspective of the ingress border router the RR sends the path to. This means that the best path will be picked based on the shortest IGP cost to the BGP next hop

my notes: RRS do not change BGP next-hop so IGP metric to iBGP next-hop is available to the RR client but the RRS can hide a path that would be best for that client  depending on topology.

>> The IOS-XR implementation allows up to three root nodes for the rSPF calculation. If you have many RR clients in one update group, then there is no need for one rSPF calculation per RR client if those RR clients will have the same policy and/or the same IGP costs to the different egress BGP border routers. This latter usually means that the RR clients are co-located (likely to be in the same POP). If that is the case, there is no need to configure each RR client as a root. The IOS-XR implementation allows to configure three, the primary, the secondary, and the tertiary root, per set of RR clients, for redundancy purposes. For the BGP ORR feature to apply to any RR client, that RR client must be configured to be part of an ORR policy group.

This feature can be considered an advanced topic that can be required only in some topologies .

This is very specific and limited in its application.

Hope to help

Giuseppe