Border Gateway Protocol, the path-vector protocol that holds the internet together and increasingly runs inside enterprise data centers. Articles tagged BGP cover fundamentals, neighbor states, path attributes, best-path selection, traffic engineering, and Cisco IOS XE configuration.
Ten routers need 45 iBGP sessions. Fifty need 1,225. Route reflectors break the iBGP split-horizon rule in a controlled way, and are the standard scaling tool for iBGP in enterprise and service provider networks.
One of the most common iBGP problems is routes that show up in the BGP table but never hit the routing table because the next-hop is unreachable. Here is when next-hop-self is the fix, and when it actually makes things worse.
iBGP has rules eBGP does not: split-horizon, full-mesh, and next-hop preservation. Ignore any of them and routes vanish silently from your network. A practical guide to configuring iBGP neighbors and full mesh.
A single mismatch in neighbor IP, AS number, or filtering keeps a BGP session down without a helpful error. A practical Cisco IOS XE eBGP configuration walkthrough including dual-homed peering and source/TTL considerations.
BGP does not auto-advertise connected interfaces like OSPF. Routes must be injected via network, redistribution, or aggregation. Understand NLRI, withdrawn routes, and the Adj-RIB pipeline to control what you advertise.
BGP evaluates 13 steps in order and stops at the first one that produces a winner. Here is the full order, with steps 1 and 2 proven on real show ip bgp output from a three-AS CML lab where the best marker moves between two paths as one attribute changes.
An unauthenticated BGP session is vulnerable to TCP-based attacks and spoofed RSTs. MD5 has been the standard for decades; TCP-AO is the modern replacement. A practical guide to BGP session authentication.
A BGP default route tells a peer to send anything it does not have a more specific path for. Different injection methods, default-originate, network, redistribute, have different behaviors and failure modes. Pick carefully.
The IPv4 BGP table has over 950,000 prefixes and growing. Route aggregation lets you advertise a summary instead of many specifics, reducing the routing table burden on your peers and the internet. Configuration and gotchas.
Weight is the first step of Cisco BGP best-path selection, Cisco-proprietary, never advertised, and local to the router. The most targeted BGP traffic engineering tool when you need one router to override AS-wide policy.
AS-path prepending is the go-to tool for influencing inbound BGP traffic: artificially lengthen the AS-path so remote ASes prefer your other exit. It is blunt, global, and has real security implications. When it works, and when it does not.
MED tries to influence how traffic enters your AS by suggesting which of your entry points the neighbor should prefer. It is optional non-transitive and easy to misuse. How MED actually works, and how to set it safely.