TCS 222 Week - 4 - Distance - Vector - and - Link - State - Routing - Algorithms
TCS 222 Week - 4 - Distance - Vector - and - Link - State - Routing - Algorithms
3
Locating and Maintaining
Routing Tables
A routing table is a database that keeps track of paths, like a
map, and uses these to determine which way to forward
traffic.
Routing decisions can be made by a central routing process
Called centralized routing
Sometimes used in connection oriented public networks
Or they can be made by a routing process in each router
Called distributed routing
Common to most connectionless network protocols
Two main methods of maintaining the routing table
Statically: manually configuring static routes
Administrator tracks topology changes
Dynamically: using a dynamic routing protocol
Routing protocol tracks topology changes
Distributed Dynamic Routing
Maintain local copy of routing table in ‘real time’
Updating routing table contents to reflect topology changes
Exchange routing information with neighbouring routers
Called routing advertisements
Algorithms (and protocols) vary
Use different updating policies
Keep different amounts of routing information (state
information)
Use different metrics to compare network paths
Distance Vector and Link State
Routing Algorithms
IP routing protocols are connectionless
Each packet is processed separately
Two categories of routing algorithm
Distance vector
Link state
Distance vector algorithms are generally considered
first-generation Distance vector Link state
routing routing
Behave acceptably in small internetworks protocols protocols
?
Basic Distance Vector Algorithms I
Periodically flood whole, or partial routing table
Some advertise to the L3 broadcast address
Stateless
Do not try to discover or handshake with neighbouring routers
Regular table routing repetition serves to confirm existence of routes
Broken routes are not advertised
Basic routing information
Use first lowest-cost route, ignoring comparable alternatives
Simple routing advertisements
Advertise only network identifier and number of hops
9
Weaknesses of Distance
Vector Routing Algorithms I
Limit network size
Cost of 16 for unreachable
networks limits network diameter
Maximum number of hops across
network
Do not support variable subnetting *
Use same address block size for all
subnets
Lowest hop count not necessarily
best path
Have no built-in security
Permits router spoofing
* Usually calledVLSM:
variable length subnet masks
Weaknesses of Distance
Vector Routing Algorithms II
Noisy protocol
Broadcasting advertisements interrupts all systems on a LAN
Routers do not maintain a neighbour table
Are stateless
Do not seek verification of arrival of routing updates from neighbours
Slow convergence (slow reaction to topology changes)
Do not necessarily advertise failure immediately
Merely omit failed networks in subsequent periodic advertisements
Takes time for all routing tables to reach agreement following a topology change
Entries for same network can differ while convergence is taking place
Susceptible to packet looping, instability and count-to-infinity
Link State Algorithms
Use short periodic Hello messages for neighbour discovery
And to confirm neighbours’ continued presence
Exchange outgoing interface cost information
Cost is dimensionless metric
Any physical value can be used, default is related to bandwidth
Exchanges used to build topological database
Complete network map
Apply Dijkstra’s shortest path first algorithm to database contents
To build shortest path first tree
Lowest cost path to each known network
Advertise topology changes immediately
Permit some degree of network structure
Referred to as areas
Are typified by OSPF and IS-IS
OSPF = Open Shortest Path First
IS-IS = Intermediate System to Intermediate System
Distance Vector vs. Link State
Algorithms I
Distance vector
Noisy protocol
Relatively high network bandwidth utilization
Simple algorithm
Low CPU overhead
Link state
Quiet protocol
Optimized for low bandwidth utilization
Complex algorithm
Potentially high CPU overhead
Distance Vector vs. Link
State Algorithms II
Feature Distance vector Link State
Convergence
(router agreement following Slow Fast
topology change)
Relies on periodic routing
Checking on neighbours Regular Hello messages
advertisements
How does each of these differ from the above RIP techniques?
Further Reading
Topics that will NOT be
Author Chapters and main topics examined from these
sections
Forouzan 21 Network Layer: Delivery,
(4th edition) Forwarding and Routing
Tanenbaum
5.2 Routing Algorithms
(4th edition) Multicast Routing
(covered in 400 Level)
Stallings
12. Routing in Switched Networks
(7th edition)