Abstract
- The inter-domain routing protocol that governs how data travels across the Internet (between ASes)
The main focus
- Scalability (ensures the global Internet stays scalable and stable, even with millions of routes.)
- Policy over pure performance (each AS uses policies to manage the traffic)
The protocol that keeps the Internet glued together, not by choosing the fastest path, but by choosing the most “policy-correct” path. It’s about scalability, stability, and business agreements between ASes.
eBGP
- Used between routers in different ASes to exchange BGP routes
- Default behaviour: routes learned from eBGP can be passed to both iBGP and other eBGP peers
iBGP
- Used between routers in the same AS to distribute BGP routes learned from eBGP across the AS
- Special rule: routes learned from one iBGP neighbour are not passed to another iBGP neighbour (to avoid loops) → This is why we need full-mesh or route reflectors
AS-Path
- Shows the list of ASes a route has passed through
- Shorter AS-Path is generally preferred, AS-Path is also one of the first attributes BGP considers when choosing a route
AS-Path for loop prevention
Helps avoid loops (if you see your own AS, drop it)
BGP Policy
- Border Gateway Protocol is not just about “fastest link”
- BGP Policies depend on business deals + reputation
Business deals + reputation
We advertise routes that pay us or maintaining reachability for our own customers and avoid carrying free transit traffic.
We also avoid unstable ASes if possible.
This mean not all possible paths are advertised and the Internet is connected enough to work, shaped by money and agreements.