Users Guide

BGP 1375
• origin
• MED
• IGP distance to the BGP next hop
Dell EMC Networking BGP does not require ECMP next hops to be in a
common AS. This behavior is enabled by default. To disable this behavior, use
the no bgp always-compare-med command.
When advertising to neighbors, BGP always advertises the single best path to
each destination prefix, even if BGP has an ECMP route to a destination.
BGP Next-Hop Resolution
BGP UPDATE messages specify a NEXT_HOP attribute for each prefix. The
NEXT_HOP attribute may be on an attached subnet for the receiver when
the UPDATE is received from an external peer. But the NEXT_HOP on
routes from internal peers or a multihop external peer is not always on a local
subnet. Thus, BGP has to resolve the BGP NEXT_HOP to one or more local
next hops (similar to how a router resolves a tunnel endpoint to a local next
hop). BGP resolves a remote NEXT_HOP by asking RTO for the longest
prefix match. As the routing table changes, the resolution for a NEXT_HOP
may change. BGP registers each remote BGP NEXT_HOP with RTO for next
hop resolution changes.
When RTO notifies BGP of a next hop resolution change, BGP finds all the
paths whose BGP NEXT_HOP is the IP address whose resolution changed
and updates the immediate next hops for each path. A next hop resolution
change triggers phase 2 of the decision process for the affected prefixes.
Dell EMC Networking allows up to 512 addresses to be registered for next-
hop resolution changes. This should be sufficient for BGP. The number of
addresses BGP needs to track is limited to the number of external peers to the
router's autonomous system (not just the external peers for the router itself)
or, if routers are configured to advertise themselves as the next hop (next-
hop-self), to the number of internal peers.
NOTE: The maximum ECMP width is limited by the chosen SDM template. All Dell
EMC Networking N3000-ON and N3100-ON Series switches can support 16-wide
ECMP when using a non-default SDM template.