Users Guide

• An SNMP walk may terminate pre-maturely if the index does not increment lexicographically. Dell EMC Networking recommends using
options to ignore such errors.
• Multiple BPG process instances are not supported. Thus, the f10BgpM2PeerInstance eld in various tables is not used to locate a peer.
• Multiple instances of the same NLRI in the BGP RIB are not supported and are set to zero in the SNMP query response.
• The f10BgpM2NlriIndex and f10BgpM2AdjRibsOutIndex elds are not used.
• Carrying MPLS labels in BGP is not supported. The f10BgpM2NlriOpaqueType and f10BgpM2NlriOpaquePointer elds are set to zero.
• 4-byte ASN is supported. The f10BgpM2AsPath4byteEntry table contains 4-byte ASN-related parameters based on the conguration.
• If a received update route matches with a local prex, then that route is discarded. This behavior results from an incorrect BGP
conguration. To overcome this issue, you can trigger a route refresh after you properly congure BGP.
• If all the IP interfaces are in non-default VRFs, then you must have at least one interface in default VRF in order to congure a routing
process that works with non-default VRFs.
Traps (notications) specied in the BGP4 MIB draft <draft-ietf-idr-bgp4–mibv2–05.txt> are not supported. Such traps
(bgpM2Established and bgpM2BackwardTransition) are supported as part of RFC 1657.
Conguration Information
The software supports BGPv4 as well as the following:
• deterministic multi-exit discriminator (MED) (default)
• a path with a missing MED is treated as worst path and assigned an MED value of (0x)
• the community format follows RFC 1998
• delayed conguration (the software at system boot reads the entire conguration le prior to sending messages to start BGP peer
sessions)
The following are not yet supported:
• auto-summarization (the default is no auto-summary)
• synchronization (the default is no synchronization)
Conguring a basic BGP network
Conguring a basic BGP network includes a few mandatory tasks and many optional tasks. The initial step is to congure a BGP routing
and BGP peers, using the address family conguration. You can also congure BGP peers using the IPv4 VRF address family.
Enabling BGP
By default, BGP is disabled on the system. Dell EMC Networking OS supports one autonomous system (AS) and assigns the AS number
(ASN). To enable the BGP process and begin exchanging information, assign an AS number and use commands in ROUTER BGP mode to
congure a BGP neighbor.
To establish BGP sessions and route trac, congure at least one BGP neighbor or peer.
In BGP, routers with an established TCP connection are called neighbors or peers. After a connection is established, the neighbors
exchange full BGP routing tables with incremental updates afterward. In addition, neighbors exchange KEEPALIVE messages to maintain
the connection.
In BGP, neighbor routers or peers can be classied as internal or external. External BGP peers must be connected physically to one another
(unless you enable the EBGP multihop feature), while internal BGP peers do not need to be directly connected. The IP address of an EBGP
neighbor is usually the IP address of the interface directly connected to the router. First, the BGP process determines if all internal BGP
peers are reachable, then it determines which peers outside the AS are reachable.
Following is the sample conguration steps to enable BGP, congure a BGP router-id and network for a router. The same congurations
have to be repeated with appropriate changes in the IP addresses for a peer or router to achieve BGP session between two devices. In the
188
Border Gateway Protocol (BGP)