Distance-Vector Routing

What algorithm is DV routing based on?

Distributed Bellman-Ford

What makes DV routing easy to implement / simple?

Each node only keeps one table with dest/cost/next-hop, doesn’t need to remember the entire path, doesn’t record alt paths.
Importantly, doesn’t require knowledge of the entire network topology like with OSPF

With DV routing, how do routers update their routing tables?

Neighboring networks (directly connected) periodically share their routing tables with each other, which then use these entries to update their own routing table

What does it mean when DV routing algorithm has converged?

All networks have the best possible routing paths to each other (table-sharing stops resulting in updates i.e tables have been optimized)

What is bouncing effect / count-to-infinity?

Bouncing effect is the slow, oscillating propagation of “bad news” / downed links, seen through slow rise in link cost per each table exchange.

Count-to-infinity describes the resulting “count upwards forever” phenomena that tables exhibit when a link goes down.

How do we solve bouncing effect / count-to-infinity? What is the limitation of these solutions?

Split Horizon: Nodes avoid advertising routes back to the same neighbor it learned that route from (stops nodes from bouncing entries back off on another)
Poison Reverse: Advertise cost of infinity if link goes down

Limitation: Only stops two-node loops

What is RIP, and some basic details about it?

RIP is the Routing Information Protocol, uses hop count as cost to destination. Max cost is 16 (=infinity), which limits size of supported network. Re-sends updateds every 30s in soft-state, with 90s being the timeout for a stale route. Send route tables upon any table changes, with 5 sec cooldown.