To solve this problem we will model the network

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: States as of early 2010. To solve this problem, we will model the network topology as a graph, a structure with nodes (vertices) connected by links (edges), as shown at the top of Figure 18-2. The nodes correspond to either switches or end points. The problem of finding paths in the network is challenging for the following reasons: 1. Distributed information: Each node only knows about its local connectivity, i.e., its immediate neighbors in the topology (and even determining that reliably needs a little bit of work, as we’ll see). The network has to come up with a way to provide network-wide connectivity starting from this distributed information. 2. Efficiency: The paths found by the network should be reasonably good; they shouldn’t be inordinately long in length, for that will increase the latency of packets. For concreteness, we will assume that links have costs (these costs could model link latency, for example), and that we are interested in finding a path between any source and destination that minimizes the total cost. We will assume that all link costs are non-negative. Another aspect of efficiency that we must pay attention to is the extra network bandwidth consumed by the network in finding good paths. 3. Failures: Links and nodes may fail and recover arbitrarily. The network should be able to find a path if one exists, without having packets get “stuck” in the network forever because of glitches. To cope with the churn caused by the failure and recovery of links and switches, as well as by new nodes and links being set up or removed, any solution to this problem must be dynamic and continually adapt to changing conditions. SECTION 18.2. ADDRESSING AND FORWARDING 3 In this description of the problem, we have used the term “network” several times while referring to the entity that solves the problem. The most common solution is for the network’s switches to collectively solve the problem of finding paths that the end points’ packets take. Although network designs where end points take a more active role in determining the paths for their packets have been proposed and are sometimes used, even those designs require...
View Full Document

This document was uploaded on 02/26/2014 for the course CS 6.02 at MIT.

Ask a homework question - tutors are online