{[ promptMessage ]}

Bookmark it

{[ promptMessage ]}

L19 - MIT 6.02 DRAFT Lecture Notes Fall 2010(Last update...

Info iconThis preview shows pages 1–3. Sign up to view the full content.

View Full Document Right Arrow Icon
MIT 6.02 DRAFT Lecture Notes Fall 2010 (Last update: November 11, 2010) Comments, questions or bug reports? Please contact [email protected] C HAPTER 19 Network Routing - II Routing Around Failures This lecture describes the mechanisms used by distributed routing protocols to handle link and node failures, packet losses (which may cause advertisements to be lost), changes in link costs, and (as in the previous lecture) new nodes and links being added to the network. We will use the term churn to refer to any changes in the network topology. Our goal is to find the best paths in the face of churn. Of particular interest will be the ability to route around failures, finding the minimum-cost working paths between any two nodes from among the set of available paths. We start by discussing what it means for a routing protocol to be correct, and define our correctness goal in the face of churn. The first step to solving the problem is to discover failures. In routing protocols, each node is responsible for discovering which of its links and corresponding nodes are still working; most routing protocols use a simple HELLO protocol for this task. Then, to handle failures, each node runs the advertisement and integra- tion steps periodically . The idea is for each node to repeatedly propagate what it knows about the network topology to its neighbors so that any changes are propagated to all the nodes in the network. These periodic messages are the key mechanism used by routing protocols to cope with changes in the network. Of course, the routing protocol has to be robust to packet losses that cause various messages to be lost; for example, one can’t use the absence of a single message to assume that a link or node has failed, for packet losses are usually far more common than actual failures. We will see that the distributed computation done in the distance-vector protocol in- teracts adversely with the periodic advertisements and causes the routing protocol to not produce correct routing tables in the face of certain kinds of failures. We will present and analyze a few different solutions that overcome these adverse interactions, which extend our distance-vector protocol. We also discuss some circumstances under which link-state protocols don’t work correctly. We conclude this chapter by comparing link-state and dis- tance vector protocols. 1
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full Document Right Arrow Icon
2 CHAPTER 19. NETWORK ROUTING - II ROUTING AROUND FAILURES 19.1 Correctness and Convergence In an ideal, correctly working routing protocol, two properties hold: 1. For any node, if the node has a route to a given destination, then there will be a usable path in the network topology from the node to the destination that traverses the link named in the route. We call this property route validity .
Background image of page 2
Image of page 3
This is the end of the preview. Sign up to access the rest of the document.

{[ snackBarMessage ]}