Link-State Routing
Which Method Is Better?: Distance-Vector vs. Link-State
When should distance-vector be used, and when should link-state
routing be used? There are arguments for and against each approach.
The choice of strategy should be based on the properties of the given
network, the main parameter being how frequently nodes are expected to
go down or come up.
Stability:
-
It is generally thought that link-state algorithms are more stable,
because each router "knows" the entire network.
-
On the other hand, if the network is highly dynamic, it is possible
for link-state algorithms to create routing loops while new topology
information is being disseminated.
Running Time:
-
Link-state algorithms converge more quickly than distance-vector
algorithms.
-
In practice, the time for convergence depends on the network graph,
the load on the routing protocol, and the exact sequence of link
failure and recovery.
Overhead:
-
Much of the overhead in link-state routing comes from the subroutine
that prevents corruption of the LSP database in case of node and/or
link failure. The distance-vector algorithm does not have this
overhead.
Memory:
-
Typically, link-state implementations require more memory than
distance-vector implementations.
Conclusion
There is no clear winner between these two strategies, especially
because for every positive feature of one protocol, a modification can
be made to the other to make it competitive again.
|
|