Skip to main content

Border Gateway Protocol (BGP)

The Border Gateway Protocol (BGP) is an inter-domain routing protocol (network protocol) used to exchange reachability information between autonomous systems on IP networks, including the public Internet.

  • Path-vector routing protocol for exchanging network reachability between autonomous systems (inter-domain routing).
  • Supports policy-based routing decisions using attributes such as AS_PATH, LOCAL_PREF, and MED (routing policy control).
  • Distinguishes between External BGP (eBGP) and Internal BGP (iBGP) for inter-AS and intra-AS route propagation (network architecture).
  • Operates over Transmission Control Protocol (TCP), providing reliable transport for routing updates on port 179 (network transport).
  • Extensible through capabilities and address family extensions (AFI/SAFI) to support IPv4, IPv6, VPNs, and other route types (routing extensibility).

More About Border Gateway Protocol (BGP)

BGP is the standardized exterior gateway protocol defined in the Internet Engineering Task Force (IETF) RFC series for exchanging routing and reachability information between autonomous systems (AS) in IP networks. It operates as a path-vector routing protocol (inter-domain routing), enabling independently administered networks to advertise prefixes, select routes, and enforce routing policies across organizational boundaries.

BGP’s primary purpose is to allow each autonomous system to control how its IP prefixes are announced to other ASes and how inbound and outbound traffic flows, based on routing policies rather than only on shortest-path metrics (routing policy control). BGP carries Network Layer Reachability Information (NLRI), describing which IP prefixes are reachable via a given AS path, along with attributes such as AS_PATH, NEXT_HOP, LOCAL_PREF, MULTI_EXIT_DISC (MED), ORIGIN, and COMMUNITY that guide route selection and propagation (route selection logic).

The protocol distinguishes between External BGP (eBGP) and Internal BGP (iBGP) sessions (network architecture). eBGP sessions operate between routers in different autonomous systems, forming the routing fabric of the Internet and many inter-organizational WANs. iBGP operates within a single AS to distribute routes learned from eBGP or other internal sources. To scale internal distribution, enterprises often use route reflectors and confederations as described in related BGP RFCs (routing scalability mechanisms).

BGP runs over TCP (network transport), using TCP port 179 for establishing sessions between peers. This reliance on TCP provides reliable delivery of BGP messages, which include OPEN, UPDATE, KEEPALIVE, and NOTIFICATION types (protocol message taxonomy). The protocol’s finite state machine governs session establishment and teardown, and timer parameters such as Hold Time control liveness detection between peers.

Through capabilities negotiation and the use of Address Family Identifier (AFI) and Subsequent Address Family Identifier (SAFI) fields, BGP supports multiple address families and service types (routing extensibility). These include unicast and multicast IPv4 and IPv6, and, when used with appropriate extensions, can support VPN-related routes and other specialized route families in provider networks.

In enterprise and institutional environments, BGP is used for multihoming to multiple service providers, interconnecting data centers, implementing complex routing policies, and integrating with Multiprotocol Label Switching (MPLS) or other carrier services (WAN and data center networking). Network engineers use BGP attributes and route filtering to manage path selection, Traffic Engineering (TE), and route aggregation, aligning routing behavior with business, performance, or resilience requirements.

Within a technical directory, BGP is categorized as an IETF-standardized inter-domain routing protocol (network-protocol/standard) used for IP reachability exchange, routing policy enforcement, and scalable control-plane coordination between autonomous systems across the Internet and private IP backbones.

Mentions