Lecture 10 - Routing Part II

Embed Size (px)

Citation preview

  • 8/2/2019 Lecture 10 - Routing Part II

    1/6

    Routing StrategiesRouting StrategiesRouting StrategiesRouting Strategies

    Fixed

    Flooding

    Random

    Adaptive

    15

    Properties of Fixed RoutingProperties of Fixed RoutingProperties of Fixed RoutingProperties of Fixed Routing

    Single permanent route for each source to destination pair

    Determine routes using a least cost algorithm

    Route fixed, at least until a change in network topology

    16

    Fixed Routing ExampleFixed Routing ExampleFixed Routing ExampleFixed Routing Example

    17

    FixedFixedFixedFixed

    RoutingRoutingRoutingRouting

    TablesTablesTablesTables

    18

  • 8/2/2019 Lecture 10 - Routing Part II

    2/6

    Properties of FloodingProperties of FloodingProperties of FloodingProperties of Flooding

    No network info required

    Packet sent by node to every neighbor

    Incoming packets retransmitted on every link except incoming link

    Eventually a number of copies will arrive at destination

    Each packet is uniquely numbered so duplicates can be discarded

    Nodes can remember packets already forwarded to keep network load in bounds

    Can include a hop count in packets

    All possible routes are tried

    Very robust

    At least one packet will have taken minimum hop count route Can be used to set up vir tual circuit

    All nodes are visited

    Useful to distribute information (e.g. routing)

    19

    FloodingFloodingFloodingFlooding

    ExampleExampleExampleExample

    20

    Properties of Random RoutingProperties of Random RoutingProperties of Random RoutingProperties of Random Routing

    Node selects one outgoing path for retransmission of incoming

    packet

    Selection can be random or round robin

    Can select outgoing path based on probability calculation

    No network info needed

    Route is typically not least cost nor minimum hop

    21

    Properties of Adaptive RoutingProperties of Adaptive RoutingProperties of Adaptive RoutingProperties of Adaptive Routing

    Used by almost all packet switching networks

    Routing decisions change as conditions on the network change

    Failure

    Congestion

    Requires info about network

    Decisions more complexTradeoff between quality of network info and overhead

    Reacting too quickly can cause oscillation

    Too slowly to be relevant

    22

  • 8/2/2019 Lecture 10 - Routing Part II

    3/6

    !

    Dynamic Routing Example Distance Vector RoutingModern computer networks tend to use dynamic routing

    algorithms.

    Distance vector routing algorithms work by having each routermaintain a table giving the best known distance to each destinationand which line to use when they get there.

    These tables are updated regularly.

    23

    Dynamic Routing Example Distance Vector RoutingRouter A computes the delay to its neighbors B and C as follows:

    B: 2msec, C: 3msec.

    Each of Routers A neighbors, in this case B and C, send delay vectorsthat show the millisecond delay from each of their machines to allrouters, as follows.Router B> A: 3, B:0, C: 7, D: 6, E: 4, F: 10

    Router C> A: 3, B:6, C: 0, D: 3, E: 12, F: 7.

    24

    A

    B

    C

    D

    F

    E

    Nodes

    Communication Link

    Dynamic Routing Example Distance Vector Routing Routing Table for Router A is then calculated

    Router A sends updates to all its neighbours.

    The measure could be anything. In this case it is delay time in milliseconds.

    Updates ripple through the network as each router does this calculation.

    Destination Link Delay

    Router A -- 0

    Router B AB 2

    Router C AC 3

    Router D AC 6

    Router E AB 6

    Router F AC 10

    25

    Key issues in this technique :

    Knowledge about the whole network

    Routing only to neighbors

    Information sharing at regular intervals

    26

  • 8/2/2019 Lecture 10 - Routing Part II

    4/6

    "

    Example Distance Vector Routing

    At start up a routers knowledge of the internetwork is sparse.Prepare a routing table with following fields:

    Network id

    Cost

    Next hop

    27 28

    Sharing of information in distance vector routing

    Routing table distribution in distance vector routing

    Updating routing table of router A

    30

  • 8/2/2019 Lecture 10 - Routing Part II

    5/6

    Final routing tablesRouter applies the following rules for updating

    process1. If the advertised station is not in the routing table, the

    router should add the advertised information to the

    table.

    2. If the advertised destination is in the routing table.

    a) If the next hop field is the same, the router should replace

    the entry in the table with the advertised one. Note that even

    if the advertised hop count is larger the advertised entry

    should replace the entry in the table because the new

    information invalidates the old.

    b) If the next hop field is not the same.i. If the advertised hop count is smaller than the one in the

    table, the router should replace the entry in the table

    with the new one.

    ii. If the advertised hop count is not smaller(same or

    larger), the router should do nothing.32

    33

    Updating process

    (Rule 2.a)(Rule 2.a)

    (Rule 1)(Rule 1)

    (Rule 2.b.i)(Rule 2.b.i)

    (Rule 2.b.ii)(Rule 2.b.ii)

    (Rule 2.b.ii)(Rule 2.b.ii)

    Adaptive RoutingAdaptive RoutingAdaptive RoutingAdaptive Routing ---- AdvantagesAdvantagesAdvantagesAdvantages

    Improved performance

    Aid congestion control

    34

  • 8/2/2019 Lecture 10 - Routing Part II

    6/6

    Two-node instability

    35

    Both A and B know

    where X is.

    Link between A and Xfails. A updates its

    table immediately.

    But before A can tell B,

    B sends its info to A!

    A, using Bs info, up-dates its table (wrongly!).

    Then A send its tableto B and B updates its

    table (again wrongly).

    Both routers keep up-dating tables, event-

    ually hitting infinity. Inthe meantime, chaos!

    Solutions to two-node instability Possible Solutions to two-node instability:

    Define infinity to be a much smaller value, such as 100. Then itdoesnt take too long to become stable. But now you cant usedistance vector routing in large networks.

    Split Horizon instead of flooding entire table to each node,only part of its table is sent. More precisely, if node B thinksthat the optimum router to reach X is via A, then B does notneed to advertise this piece of info to A the info has alreadycome from A.

    Split Horizon and Poison Reverse Normally, the distance

    vector protocol uses a timer. If there is no news about a route,the node deletes the route from its table. So when A neverhears from B about the route to X, it deletes it. Instead, Node Bstill advertises the value for X, but if the source of info is A, itreplaces the distance with infinity, saying Do not use thisvalue; what I know about this route comes from you.

    36

    Three-node instability - no solution here!?

    37