רמי פוזיס

Flexibility and extendibility of Software Defined Networks allows development of diverse network management and flow monitoring techniques. Yet, there are inherent tradeoffs between the quality of flow monitoring and the required network resources. In particular, collecting flow statistics, at the level of specific source-destination addresses (and, moreover, specific protocols and ports), requires too many flow table entries. This problem is emphasized by the difficulty of anticipating the individual flows that need to be monitored. In this paper we propose a method for dynamic flow discovery at any required spatial resolution. In addition, we propose a method for balancing the monitoring effort among the switches. These methods allow increasing the spatial resolution of traffic monitoring with minimal effects of the network performance.