Management of routing protocols in ip networks
Aman Shakil Shaikh, Anujan Varma · 2003
The Internet connects millions of computers world-wide and allows them to communicate with one another. The underlying routing infrastructure, consisting of special purpose computers called routers and links interconnecting them, makes this communication possible by delivering data from its sources to its destinations. Effective management of the routing infrastructure—configuring the routers, identifying and troubleshooting problems, performing maintenance to fix problems, and upgrading the infrastructure to support new services—is vital to the smooth functioning of the Internet. Performing these management tasks is extremely challenging because the infrastructure is a large distributed system having a high degree of dynamism. Despite this, the management of the routing infrastructure has not received much research attention. In this dissertation, we focus on the management of OSPF (Open Shortest Path First), one of the widely run protocols on the routers. In particular, we focus on three aspects of OSPF management: monitoring, characterization of its dynamics, and maintenance. Very few tools exist for monitoring OSPF behavior effectively. In this dissertation, we propose a scalable, effective and passive monitoring system for OSPF. The highlights of our monitoring system are: (a) capturing OSPF routing messages, while remaining totally passive to the network and visible only at the point of attachment; (b) an efficient methodology for online generation of alerts that operators can use to identify and troubleshoot problems; and (c) sages, and propose a classification of these messages into three categories: (a) messages triggered by OSPF's soft-state refresh, (b) messages triggered by events that change the status of the network, and (c) “duplicate” messages received due to the redundancy in OSPF's reliable message flooding mechanism. This classification allows us to determine the baseline rate of OSPF messages, network problems and protocol overheads. The dissertation presents a case study of applying this classification to one month's worth of OSPF routing messages collected from an enterprise network. Operators running large networks frequently upgrade routing hardware and software in order to fix problems and to introduce new features. While performing such maintenance tasks, the operators often must withdraw the router undergoing maintenance from forwarding, resulting in disruption to the data traffic. (Abstract shortened by UMI.)