Monitoring Flow Aggregates with Controllable Accuracy
2007 (English)In: Lecture Notes in Computer Science, ISSN 0302-9743, E-ISSN 1611-3349, Vol. 4787, 64-75 p.Article in journal (Refereed) Published
In this paper, we show the feasibility of real-time flow monitoringwith controllable accuracy in today’s IP networks. Our approach is based onNetflow and A-GAP. A-GAP is a protocol for continuous monitoring ofnetwork state variables, which are computed from device metrics usingaggregation functions, such as SUM, AVERAGE and MAX. A-GAP isdesigned to achieve a given monitoring accuracy with minimal overhead. AGAPis decentralized and asynchronous to achieve robustness and scalability.The protocol incrementally computes aggregation functions inside the networkand, based on a stochastic model, it dynamically configures local filters thatcontrol the overhead and accuracy. We evaluate a prototype in a testbed of 16commercial routers and provide measurements from a scenario where theprotocol continuously estimates the total number of FTP flows in the network.Local flow metrics are read out from Netflow buffers and aggregated in realtime.We evaluate the prototype for the following criteria. First, the ability toeffectively control the trade off between monitoring accuracy and processingoverhead; second, the ability to accurately predict the distribution of theestimation error ; third, the impact of a sudden change in topology on theperformance of the protocol. The testbed measurements are consistent withsimulation studies we performed for different topologies and network sizes,which proves the feasibility of the protocol design, and, more generally, thefeasibility of effective and efficient real-time flow monitoring in large networkenvironments.
Place, publisher, year, edition, pages
2007. Vol. 4787, 64-75 p.
Computer simulation; Error analysis; Network protocols; Robustness (control systems); Routers; Scalability; Controllable accuracy; Estimation error; Local flow metrics
Engineering and Technology
IdentifiersURN: urn:nbn:se:kth:diva-9465DOI: 10.1007/978-3-540-75869-3_6ISI: 000251162800006ScopusID: 2-s2.0-38149116163OAI: oai:DiVA.org:kth-9465DiVA: diva2:114072
Conference: 10th IFIP/IEEE International Conference on Management of Multimedia and Mobile Networks and Services. San Jose, CA. OCT 31-NOV 02, 2007 2008-11-052008-11-052011-09-12Bibliographically approved