BGP

Definition

BGP, the Border Gateway Protocol, is the routing protocol that exchanges reachability information between autonomous systems on the internet, determining the paths data takes between different networks operated by internet service providers, cloud platforms, and large organizations. It is the protocol that effectively holds the global internet together by letting independently operated networks agree on how to route traffic to each other.

Overview

BGP exists because the internet is not one network but a collection of thousands of independently administered networks, called autonomous systems, each identified by a unique autonomous system number, that must agree on how to forward traffic destined for addresses outside their own network. Without a shared protocol for exchanging this reachability information, no network operator could know which path to send traffic on to reach a destination hosted by a different provider anywhere else in the world. Mechanically, BGP routers exchange information about which IP address ranges, or prefixes, they can reach and through which path of autonomous systems, forming what is called an AS path. When a router learns of multiple possible paths to the same destination, it applies a defined decision process — considering factors like path length, local policy preferences, and administrative weights — to select a single best path, which it then advertises onward to its own neighbors. BGP is a path-vector protocol, distinguishing it from interior routing protocols like OSPF, which compute shortest paths within a single organization's network using link-state information rather than negotiating policy between independently operated networks. BGP operates specifically between autonomous systems, which is why it is called an exterior gateway protocol, in contrast to interior gateway protocols such as OSPF that manage routing within a single autonomous system. Cloud providers, internet service providers, and any organization large enough to operate its own autonomous system use BGP to control how their address space is reached from the rest of the internet and to choose among multiple upstream providers for outbound traffic, a practice known as multihoming. In practice, BGP underlies nearly every cross-network path that internet traffic takes, is used by cloud platforms to advertise service address ranges globally, and is configured by network engineers at ISPs and large enterprises to control traffic engineering, failover between multiple internet connections, and defense against certain classes of routing attacks. The well-known limitation is that BGP was designed assuming a level of mutual trust between network operators that does not fully hold today: because BGP has no built-in cryptographic verification that an autonomous system is authorized to advertise a given prefix, misconfigurations or malicious announcements, called route hijacks, can redirect traffic through unintended paths, which is why supplementary technologies like Resource Public Key Infrastructure have been developed to add cryptographic route-origin verification on top of BGP rather than replacing the underlying protocol outright.

Specification

• Path-vector protocol exchanging reachability between autonomous systems
• Exterior gateway protocol operating between, not within, networks
• Best-path selection based on AS path length and configurable policy
• Enables multihoming across multiple upstream internet providers
• Used by cloud providers to advertise service address ranges globally
• No built-in cryptographic verification of route announcements
• Foundation protocol for global internet interconnection and routing

Use Cases

Alternatives

Frequently Asked Questions

Related in DevOps