Outline: Routing

• Routing Problem
• Source Routing
• Distance Vector
• Link State
• Broadcast

Routing Problem

• in an internet, data has to somehow be forwarded from the source, through a number of intermediate systems, to the destination
• some internets that require routing:
  • Internet: datagrams, routing occurs for each datagram
  • ATM: virtual circuits, routing occurs at connection setup time
  • 1980's email: source route described path across sequence of (occasionally) interconnected hosts
• Routing problem:
  • who is responsible for determining the path from the source to the destination?
  • using what algorithm?

Routing Ideas

• minimum: each switch must know what port an item of data must be sent on
• ports connect to topological neighbors
• maximum: each switch and host has a map of the entire network, could do source routing
• broadcast: why not simply send data everywhere?
  • doesn't scale
  • doesn't handle loops
  • instead, "broadcast" routing information
• how fast does the routing information change?

Distance Vector Routing

• each neighbor tells me how far it is from every host it knows about
• if a neighbor N has a shorter path than I do to host X, I will route packets or circuits to X through N
• in case of ties, I pick one at random.
• "broadcasting": distance information about the whole network eventually reaches the entire network

Distance Vector Algorithm

1. keep a routing table indexed by destination:
   • distance
   • port (vector)
2. send to each neighbor both periodically and when my routing table changes
3. each message contains destination, distance from my routing table
4. when receiving a message on port P:
   • add one (or link metric) to distance of each entry in the message
   • replace entries in the table with entries from message if either:
     • distance message < distance table, or
     • port table = P

Distance Vector example

• My routing table:

  Destination	Distance	Port
  A	4	1
  B	2	4
  X	5	2

• message on port 4: (A, 2), (B, 3), (X, 5)
• New routing table:

  Destination	Distance	Port
  A	3	4
  B	4	4
  X	5	2

Link-State Routing

• each node keeps a table of its neighbors, with destination and metric
• each node broadcasts its neighborhood table whenever it changes
• from sequence of broadcasts each node can figure out the network topology
• node uses private map to route packets
• every switch must use equivalent algorithm, or we use source routing

Link-State -- Distance Vector

• LS routing allows the exchange of much more precise information than DV does (DV exchanges a single "distance" metric).
• LS must exchange more information overall, and each host must compute a route based on the resulting topology information. With DV, computation of the route is automatic.
• LS requires some notion of the "age" of information. A reliable DV also needs a way to make information obsolete, since otherwise information takes a long time to become obsolete.

Broadcasting:
Flooding Algorithm

1. Keep track of all packets received
2. On receipt of a packet from port i:
   • If you have received it before, discard it.
   • Otherwise, forward it on all ports except port i.

More Practical Flooding

• each sender numbers all broadcast packets sequentially
• each node only needs to remember the most recent sequence number from each sender
• only works if there is no danger of re-ordering.

Another Flooding Algorithm

• if:
  • you have Link-State information, and
  • all switches are running the same algorithm
  can build spanning tree back to the sender of a broadcast message
• only forward the message to the nodes that are "below" you on this spanning tree.
• no need for sequence numbers, "remembering"