We cannot assume idealized routing in practice:

• Routers may not be identical
• Network is not “flat”

In terms of scale, there are billions of potential destinations.

• It is impossible to store all possible destinations in routing tables - routing table exchange would swamp links

Instead, we have an administrative autonomy in practice:

• Internet is a network of networks
• ISP will have its own autonomous system
  • In every autonomous system, run a particular routing algorithm based on some metrics
• Each network admin may want to control routing in its own network
• LS/DV algorithms run in the scope of a particular network
• To route outside of a network, use inter routing algorithm in addition to intra-ISP routing

For scalable routing, routers are aggregated into regions known as “autonomous systems” (AS) - aka domains.

• Intra-AS (Intra-domain) involves routing within the same AS (”network”)
  • All routers in an AS must run the same intra-domain protocol
  • Routers in different AS can run different intra-domain routing protocols
  • AS have a gateway router
    • Exists at “edge” of own AS ⇒ link routers into other AS’es
• Inter-AS (Inter-domain) involves routing among AS’es
  • Gateway routers perform inter-domain routing in addition to intra-domain routing

In interconnected ASes, the forwarding table is configred by intra- and inter-AS routing algorithms.

• Intra-AS routing ⇒ determine entries for destinations within AS
• Inter-AS & intra-AS ⇒ determine entries for external destinations