Border Gateway Protocol

Border Gateway Protocol
Johannes Rössel · CC BY-SA 3.0 · source
Name	Border Gateway Protocol
Acronym	BGP
Developer	Internet Engineering Task Force (IETF)
Initial release	1989
Latest release	BGP-4 (RFC 4271)
License	Open standards
Category	Exterior gateway protocol
Website	Internet Engineering Task Force

Border Gateway Protocol

Border Gateway Protocol is the principal exterior gateway protocol used to exchange routing information between autonomous systems on the Internet. It enables disparate networks operated by organizations such as Cisco Systems, Juniper Networks, Google, Amazon (company), and Microsoft to interconnect at Internet exchange points like DE-CIX, LINX, and AMS-IX. BGP sits at the interdomain layer of the global routing system and is specified and evolved within the Internet Engineering Task Force through RFCs.

Overview

BGP is an inter-autonomous system routing protocol standardized by the Internet Engineering Task Force and deployed by network operators including Akamai Technologies, Facebook, Cloudflare, Verizon Communications, and AT&T. It operates between routers manufactured by vendors such as Cisco Systems, Juniper Networks, Huawei Technologies, and Arista Networks and is implemented in routing platforms like Quagga, FRRouting, BIRD (routing daemon), and proprietary systems from Cisco IOS. Large-scale deployments occur at peering points such as Equinix, exchanges like DE-CIX, and backbone networks run by Level 3 Communications and NTT Communications.

History and Development

BGP was developed in the late 1980s and documented in early standards work by the Internet Engineering Task Force with contributions from organizations like MCI Communications and research groups at universities such as Stanford University and MIT. Its first operational use followed earlier inter-domain approaches embodied by protocols used within the ARPANET and the transition to commercial networks involving entities like UUNET and BBN Technologies. Major revisions culminating in BGP-4 were driven by the need to support classless inter-domain routing (CIDR) to address exhaustion issues noted alongside reports from bodies such as the Internet Architecture Board and standards published as RFCs through IETF Working Group processes.

Protocol Operation and Concepts

BGP peers form sessions over TCP between routers designated as autonomous system (AS) nodes administered by organizations like AT&T, Sprint Corporation, Deutsche Telekom, and Orange S.A.. Route advertisements include path attributes such as AS_PATH, NEXT_HOP, LOCAL_PREF, and MED; these attributes are processed by implementations on platforms like Cisco IOS XR, Juniper Junos, and open-source daemons originating from projects at University of Zagreb and others. Session establishment involves TCP and capabilities negotiation; route propagation follows policy controls used by network engineering teams at companies such as Facebook and Google Fiber operating within data centers at locations like Silicon Valley and cloud regions operated by Amazon Web Services.

Route Selection and Policies

BGP route selection follows ordered tie-breaking rules prioritized by attributes including LOCAL_PREF, AS_PATH length, origin type, MED, and IGP cost to NEXT_HOP, with operators from corporations such as IBM, Oracle Corporation, SAP SE, and research networks like Internet2 shaping policies. Routing policies implement export and import filters using route maps and prefix-lists on routers from Cisco Systems and Juniper Networks; they are coordinated across peers at exchanges such as AMS-IX and transit providers including Tata Communications and CenturyLink (Level 3).

Security and Vulnerabilities

BGP has been subject to incidents including prefix hijacking and route leaks that affected organizations like Google, Amazon (company), Facebook, and national networks such as those in Ukraine and Pakistan; high-profile events prompted attention from entities like the National Institute of Standards and Technology and the Internet Society. Security mitigations and mechanisms include RPKI developed with input from ARIN, RIPE NCC, and APNIC; BGPsec and route filtering proposals have been advanced through the IETF and evaluated by vendors such as Cisco Systems and research teams at Carnegie Mellon University, University of California, Berkeley, and Princeton University. Operational best practices promoted by organizations like MANRS aim to reduce misconfiguration risks impacting content providers such as Netflix and Dropbox.

Implementations and Deployment

BGP implementations are available from vendors including Cisco Systems, Juniper Networks, Huawei Technologies, Arista Networks and open-source projects like Quagga, FRRouting, and BIRD (routing daemon). Cloud providers including Amazon Web Services, Microsoft Azure, and Google Cloud Platform offer BGP-based VPN and peering services integrated with edge devices and virtual routers. Academic and municipal networks such as Internet2, GEANT, and city-wide initiatives in New York City and London run BGP across campus and metropolitan exchanges, often coordinated with carriers like Telefonica and Vodafone.

Extensions and Future Directions

Future directions for BGP include stronger cryptographic validation (BGPsec), enhanced prefix origin validation via RPKI supported by registries like ARIN and RIPE NCC, and automation approaches leveraging standards from the IETF and orchestration platforms from VMware and Red Hat. Research labs at MIT, Stanford University, ETH Zurich, and companies such as Cisco Systems and Google explore integration with software-defined networking initiatives like OpenFlow and network telemetry advances from projects at CAIDA and RIPE NCC. Emerging operational models involve distributed edge architectures adopted by providers such as Cloudflare and content delivery networks like Akamai Technologies.

Category:Internet protocols