---

Link failure processing by the Distance Vector Algorithm

---

• Example Distance Vector computation when a link fails
  • This is the final result from the complete distance vector computation from the previous class notes:

    Notes:

    The values inside the red circles are: the minimum cost (distance) to a destination node Example: in F's distance table minimum cost path (from F) to A is 1 (via A) minimum cost path (from F) to B is 2 (via A)        minimum cost path (from F) to D is 2 (via G) ...

    ---

    ---

    ---

    ---

  • Suppose the link FG fails:

    Fact:

    Node F and node G will detect the link failure (and set the link cost to infinity

    Dictactical node:

    I will show the computation done by node F only          (Computation by node G is similar)

  ---

  ---

  ---

  ---

  ---

  ---

• Compuation by Distance Vector Algorithm for a link failure
  • After node F updated the link cost FG to infinity:

    Node F detects that the miminum distance to node G has changed: Before: min = min(1,3) = 1 After: min = min(∞,3) = 3

    Therefore:

    Node F will transmit the distance vector (F,G,3) to all its neighbors (BTW, G will not receive the distance vector update because the link FG has failed !!!!)

    Graphically:

    ---

    ---

    ---

    ---

  • STOP !!

    Look at the routing tables : A packet from node B sent to node G will be routed as follows: Infinite loop: B   ->   A   ->   F   ->   A   ->   F   ->   .....

    ---

    ---

  • Conclussion:

    There is a routing loop in the network !!! (Loops are one of the main problems with Distance Vector routing)

  • Fortunately, the routing loop is temporally in nature.

    It will be resolved when update messages are propagated to the rest of the network

    ---

    ---

    ---

    ---

    ---

  • We continue with the example....

    When node A receives F's update message, it will then update its distance table as follows:

  • There is a distance change in A's distance table, namely:

    Before: After: To Cost Cost To Cost Cost via C via F via C via F -------------------- ----------------------- G 3 2 G 3 4

    (The minimum distance changed from 2 to 3 ).

    Result:

    Node A sends (A, G, 3) to all its neighbors

    Graphically:

    Notes:

    The update message (A, G, 3) are received by neighbors B, C, E and F. Each neighbor will update its distance table: The updates by nodes B, C, E and F did not cause the route cost to change !!!! When the (minumum) route cost does not change: There is no need to send any distance vector messages !!!!

    ---

    ---

  • Just to show you why there is no need to send a distance vecor update message:

    Suppose that node B does send the distance vector update message (B, G, 3) to its neighbors: Result: B's neighbor nodes will compute the same distance values !!! (so why send it ???) !!!

    ---

    ---

    ---

  • Result:

    The Distance Vector Algorithm has converged Note: Node G's message was not incorporated !!!! (It's processed in a similar fashion) According to the routing table values, the Distance Vector Algorithm has found the follow path to the destination G:

    Thus avoiding the failed link FG !!!!