Week 6– CST– 311 Intro to Computer Networks

Module 6 – The Network Layer- Control Plane

This week we learned about the second part of the Network Layer, the control plane. The control plane is where routing takes place. There are two ways of structuring the control plane, pre-router control and logically centralized control. With pre-router control, each router has its own routing algorithm, and they interact with each other to create forwarding tables. In logically centralized control, a controller interacts with local agents and computes the forwarding tables for the routers.

In section 5.2 we covered routing algorithms. There are centralized routing algorithms, which compute the least-cost path from source to destination using full knowledge of the network. These algorithms are called link-state or LS algorithms because the algorithms are aware of the cost of each link in the network. Dijkstra’s algorithm is an algorithm that computes the least-cost path from one node to all other nodes in the network. Our text explains that Dijkstra’s algorithm is iterative and “after the kth iteration, the least-cost paths are known to k destinations, and among the least-cost paths to all destinations nodes, these k paths will have the k smallest costs.” Although LS algorithms are effective, they can run into oscillation issues when the links are not symmetrical. What our text suggests as a solution to this issue is to ensure that not all routers run the LS algorithms at the same time.

The other type of routing algorithms are decentralized algorithms. Here the routers determine the least-cost path by iteration. Each node begins with knowledge of only the cost of its own directly connected links. Through iteration and exchange of information, with its neighbor nodes, each node calculates the least-cost path to a destination. The algorithm we studied was the distance-vector or DV algorithm. The name comes from the fact that each node has a vector of costs to all other nodes in the network. The DV algorithm can also run into an issue called count-to-infinity which can happen when the value of a link increases. Again, the text suggests a solution called poisoned reverse which has one node lie to another node and advertise that one of its links has a distance of infinity to keep it from using that route. Unfortunately, loops with count-to-infinity that have 3 or more nodes will not be detected by poisoned reverse.

Comments

Popular posts from this blog