Efficient Network and Traffic Monitoring

Summary form only given. Offering reliable novel services in modern heterogeneous networks is a key challenge and an important prospective income source for many network operators and providers. Providing reliable future services in a cost effective scalable manner requires efficient use of networking and computation resources. This can be done by making the network more self-enabled, i.e. making it capable of making distributed local decisions regarding the utilization of the available resource. However, such decisions must be correlated in order to achieve a global overall goal (maximum utilization or maximum profit, for example). A key building block for all such systems is the ability to monitor the network parameters and the relevant traffic, and to infer from these measurements the relevant information needed in each one of the local decision points. Due to the heterogeneous nature of modern networks and to the very high amount of traffic, even monitoring a local location introduces significant difficulties. It is much more challenging to decide what type of traffic or network information should be collected at each network segment in order to acquire the needed global information without investing too much effort in the monitoring process or its management. In fact, efficient network and traffic monitoring may become a very significant ingredient in the ability to provide modern network services in a cost effective way. This tutorial deals with practical and efficient techniques to retrieve information from modern network devices. We start by examining the SNMP suit and the various methods to collect information from possibly large MIB tables. Then we develop a framework for quantifying resource (bandwidth and CPU) utilization in distributed network management. To demonstrate the practical impact of this framework, advanced techniques for efficient reactive traffic monitoring, efficient QoS parameter monitoring, and multimedia application monitoring, together- - with empirical results showing the overhead reduction will be presented. The tutorial continues with an example for a reliable, efficiency aware monitoring system that combines the above techniques with the SNMP framework, and time allowing novel techniques for efficient statistical monitoring