Autonomous Systems

The Foundation of BGP

Autonomous Systems (AS) are the fundamental building blocks of BGP and the Internet. Understanding AS concepts is crucial for BGP mastery.

What is an Autonomous System?

An Autonomous System (AS) is a collection of IP networks and routers under the control of a single administrative entity that presents a common, clearly defined routing policy to the Internet.

Key Characteristics

  • Single Administration: Managed by one organization
  • Unified Policy: Common routing policy to external networks
  • Unique Identifier: Each AS has a unique AS number
  • Internal Connectivity: All networks within the AS are reachable

AS Number (ASN) Allocation

AS numbers are managed by Regional Internet Registries (RIRs) and assigned to organizations that need to connect to the Internet with their own routing policy.

AS Number Range Type Usage Availability
1 - 64511 16-bit Public Internet routing Limited, carefully allocated
64512 - 65534 16-bit Private Internal use only Free for private use
65535 Reserved Special use Not available
65536 - 4199999999 32-bit Public Internet routing Abundant, modern standard
4200000000 - 4294967294 32-bit Private Internal use only Free for private use

Types of Autonomous Systems

Stub AS

Characteristics:

  • Single connection to Internet
  • No transit traffic
  • Default route only

Example: Small enterprise

Multi-homed AS

Characteristics:

  • Multiple ISP connections
  • No transit traffic
  • Receives partial/full routes

Example: Large enterprise

Transit AS

Characteristics:

  • Multiple connections
  • Carries transit traffic
  • Full Internet routes

Example: ISP

Internet Hierarchy

The Internet is structured in a hierarchical manner with different tiers of ISPs:

Internet Hierarchy Structure

Tier 1 ISPs
AT&T (AS 7018)
Verizon (AS 701)
Level3 (AS 3356)


Tier 2 ISPs
Regional ISP (AS 6461)
Regional ISP (AS 3257)


Tier 3 ISPs / Enterprises
Local ISP (AS 65001)
Enterprise (AS 65002)

Tier Classification

Tier 1

Characteristics:

  • Settlement-free peering
  • No upstream providers
  • Global reach
  • Large backbone networks

Count: ~12-15 worldwide

Tier 2

Characteristics:

  • Purchase transit from Tier 1
  • Peer with other Tier 2
  • Regional/national coverage
  • Mix of peering and transit

Count: Hundreds worldwide

Tier 3

Characteristics:

  • Purchase all transit
  • Local/regional focus
  • Serve end customers
  • Limited peering

Count: Thousands worldwide

AS Relationships

Different types of relationships exist between autonomous systems:

Relationship Description Traffic Flow Financial Model
Customer-Provider Customer pays provider for Internet access Customer → Provider → Internet Customer pays provider
Peer-to-Peer Mutual exchange of traffic Bidirectional, balanced Settlement-free
Sibling Same organization, different ASes Free exchange No charges

BGP Communities and AS Relationships

BGP communities are often used to signal AS relationships and implement policies:

Common Community Values

  • No-Export (65535:65281): Don't advertise to eBGP peers
  • No-Advertise (65535:65282): Don't advertise to any peer
  • Local-AS (65535:65283): Don't advertise outside local AS
  • Custom Communities: Provider-specific policy tags

AS Path Attributes

The AS path is fundamental to BGP operation and shows the sequence of ASes a route has traversed:

AS Path Examples

# Simple AS path
Network: 192.0.2.0/24
AS Path: 65001 65002 65003
Origin: AS 65003

# AS path with prepending
Network: 192.0.2.0/24
AS Path: 65001 65002 65002 65002 65003
Origin: AS 65003 (prepended for traffic engineering)

# AS path with AS set
Network: 192.0.2.0/24
AS Path: 65001 {65002,65003} 65004
Origin: Aggregated from multiple sources

Private AS Numbers

Private AS Considerations

Private AS numbers should be removed before advertising to the Internet:

  • 16-bit Private: 64512-65534
  • 32-bit Private: 4200000000-4294967294
  • Removal: Use "remove-private-as" command
  • Replacement: ISP AS number replaces private AS
Previous: Key Terminology Next: BGP Speakers

BGP Speakers

The Participants in BGP

BGP speakers are the routers that participate in BGP routing. Understanding their roles and relationships is essential for BGP network design.

What is a BGP Speaker?

A BGP speaker is a router that implements the BGP protocol and can exchange routing information with other BGP speakers. It maintains BGP sessions, processes routing updates, and makes routing decisions based on BGP attributes.

Core Functions

Session Management

Establish and maintain BGP sessions with neighbors using TCP connections.

Route Processing

Receive, process, and advertise routing information according to configured policies.

Policy Implementation

Apply routing policies for import, export, and path selection decisions.

Path Selection

Choose the best path to destinations using BGP's path selection algorithm.

Types of BGP Speakers

Based on AS Relationship

eBGP Speaker

Definition: BGP speaker that peers with routers in different autonomous systems

Characteristics:

  • Usually directly connected
  • TTL = 1 by default
  • Administrative distance = 20
  • Next-hop is typically changed
iBGP Speaker

Definition: BGP speaker that peers with routers in the same autonomous system

Characteristics:

  • May be multi-hop
  • TTL = 255 by default
  • Administrative distance = 200
  • Next-hop is typically preserved

BGP Speaker Configuration

Basic BGP Speaker Setup

Basic Configuration

# Configure BGP speaker
router bgp 65001
 bgp router-id 1.1.1.1
 bgp log-neighbor-changes
 
 # eBGP neighbor
 neighbor 203.0.113.1 remote-as 65002
 neighbor 203.0.113.1 description "eBGP to ISP"
 
 # iBGP neighbor
 neighbor 192.168.1.2 remote-as 65001
 neighbor 192.168.1.2 description "iBGP peer"
 neighbor 192.168.1.2 update-source loopback0
 neighbor 192.168.1.2 next-hop-self
Previous: Autonomous Systems Next: Session Types

BGP Session Types

Understanding BGP Sessions

BGP sessions are the foundation of BGP communication. Understanding different session types and their characteristics is crucial for proper BGP implementation.

BGP Session Fundamentals

A BGP session is a TCP connection between two BGP speakers that exchange routing information. Each session is characterized by its relationship type, configuration, and operational parameters.

Session Establishment Process

TCP Connection
→
BGP Open
→
Keepalive Exchange
→
Route Exchange

External BGP (eBGP) Sessions

eBGP sessions connect BGP speakers in different autonomous systems. These sessions are typically used for inter-domain routing.

eBGP Characteristics
  • AS Numbers: Different for each peer
  • Administrative Distance: 20
  • TTL: 1 (directly connected)
  • Next-hop: Changed to advertising router
  • AS Path: Prepended with local AS
eBGP Configuration 
router bgp 65001
 neighbor 203.0.113.1 remote-as 65002
 neighbor 203.0.113.1 description "eBGP to ISP"
 neighbor 203.0.113.1 password secretkey
 network 192.168.1.0 mask 255.255.255.0

Internal BGP (iBGP) Sessions

iBGP sessions connect BGP speakers within the same autonomous system. These sessions distribute externally learned routes throughout the AS.

iBGP Characteristics
  • AS Numbers: Same for all peers
  • Administrative Distance: 200
  • TTL: 255 (may be multi-hop)
  • Next-hop: Typically preserved
  • AS Path: Not modified
iBGP Rules
  • Split Horizon: Routes from iBGP not advertised to iBGP
  • Synchronization: iBGP routes need IGP support
  • Next-hop Reachability: Must be reachable via IGP
  • Full Mesh: Traditionally required
Previous: BGP Speakers Next: Path Vector Algorithm

Path Vector Algorithm

The Heart of BGP

The path vector algorithm is what makes BGP unique among routing protocols. It maintains complete path information to prevent loops and enable sophisticated policy control.

What is Path Vector?

Path vector is a routing algorithm that maintains the complete path to each destination. Unlike distance vector protocols that only know the distance and next hop, path vector protocols maintain the entire sequence of autonomous systems a route has traversed.

Key Concepts

Path Information

Each route includes the complete AS path, showing every AS the route has passed through.

AS Path: 65001 65002 65003

Loop Prevention

If a router sees its own AS in the path, it rejects the route, preventing loops.

Rejected: 65001 65002 65001

Algorithm Comparison

Algorithm Information Maintained Loop Prevention Scalability Policy Control
Distance Vector Distance + Next Hop Split Horizon, Poison Reverse Limited Minimal
Link State Complete Topology SPF Algorithm Good Limited
Path Vector Complete Path AS Path Loop Detection Excellent Extensive
Previous: Session Types Next: BGP States

BGP States

BGP Session Lifecycle

BGP sessions progress through a defined set of states during establishment and maintenance. Understanding these states is crucial for troubleshooting BGP issues.

BGP Finite State Machine

BGP operates as a finite state machine (FSM) with six defined states. Each state represents a specific phase in the BGP session lifecycle.

Idle
→
Connect
→
Active
→
OpenSent
→
OpenConfirm
→
Established

BGP State Descriptions

1. Idle State

Idle State

Description: Initial state where BGP waits for a start event.

Characteristics:

  • No resources allocated
  • No TCP connection attempted
  • Waiting for manual or automatic start
  • ConnectRetry timer not running

6. Established State

Established State

Description: BGP session fully established, exchanging routes.

Characteristics:

  • Session fully operational
  • Route exchange in progress
  • Keepalive messages sent periodically
  • Hold timer reset on each message

BGP Timers

Various timers control BGP state transitions:

Timer Default Value Purpose States Used
ConnectRetry Timer 120 seconds Time between connection attempts Connect, Active
Hold Timer 180 seconds Maximum time between messages OpenSent, OpenConfirm, Established
Keepalive Timer 60 seconds Keepalive transmission interval OpenConfirm, Established
Previous: Path Vector Algorithm Next: BGP Header Format