This paper presents a new publicly available dataset from GÉANT, the European Research and Educational Network. This dataset consists of traffic matrices built using full IGP routing information, sampled Netflow data and BGP routing information of the GÉANT network, one per 15 minutes interval for several months. Potential benefits of publicly available traffic matrices comprise improving our understanding of real traffic matrices, their dynamics, and to make possible the benchmarking of intradomain traffic engineering methods. Categories and Subject Descriptors MOTIVATIONA lot of effort has been put the last few years on trying to infer traffic matrices based on SNMP link counts [1,2,3,4]. The approach of relying on the raw traffic demand [5,6] is rarely used as the burden of the measurement and storage infrastructure is significant [7]. Still, recent works [8,9] indicate that obtaining precise traffic matrices is not out of reach.Contrary to single capture points traffic traces [10,11,12] or BGP routing data [13,14] for which numerous publicly available datasets exist, publicly available traffic matrices coming from a real network are rare. The only publicly available set of traffic matrices to our knowledge is at http://www.cs.utexas.edu/ yzhang/research/AbileneTM/ based on data from the Abilene network. Developing intradomain traffic engineering tools or traffic matrix modeling require real datasets to validate the tools or the models. Without publicly available datasets, no comparisons with alternative techniques or models can be performed. To contribute to filling this lack in the networking community, this paper presents a publicly available dataset consisting of intradomain THE GÉANT NETWORKGÉANT is the pan-European research network and it is operated by DANTE. It carries research traffic from the European National Research and Education Networks (NRENs) connecting universities and research institutions. GÉANT has a PoP in each European country 3 . All the routers of GÉANT are border routers. GÉANT is composed of 23 routers interconnected using 38 links. In addition, GÉANT has 53 links with other domains. GÉANT uses ISIS to compute its intradomain routes. The IGP weights of GÉANT are mainly based on the inverse of the link capacities with some manual tunings. We obtained a libpcap trace of ISIS for the purpose of building a model of the GÉANT topology.In order to build an accurate model of GÉANT suitable for the computation of its intradomain traffic matrices, we also obtained from DANTE the interdomain routes known by GÉANT as well as a trace of the traffic transiting accross GÉANT [15]. The interdomain routes are obtained from BGP and the traffic trace is collected using Netflow. We describe these two datasets in the following paragraphs. BGP Routing dataIn GÉANT, the BGP routes are collected using a dedicated workstation running GNU Zebra [16], a software implementation of different routing protocols including BGP. The workstation has an iBGP session with all the border routers of the networ...

Abstract-We explain how the TOTEM toolbox can be used to engineer an operational network. TOTEM is an open source TOolbox for Traffic Engineering Methods which covers IPbased and MPLS-based intradomain traffic engineering (TE) algorithms, but also interdomain TE. In this paper, we use the toolbox as an off-line simulator to optimise the traffic of an operational network. To help an operator to choose between an IP-based or MPLS-based solution, or to find the best way to loadbalance a network for a given traffic, our case study compares several IP and MPLS routing algorithms, evaluates the impact of hot-potato routing on the intradomain traffic matrix, and analyses the worst-case link failure. This study reveals the power of a toolbox that federates many traffic engineering algorithms.

We compare and evaluate different methods to infer groups of correlated failures. These methods try to group failure events occurring nearly simultaneously in clusters. Indeed if several failures occur nearly at the same moment in a network, it is possible that these failures have the same root cause. The input data of our algorithms are IP failure notifications that can be provided by several sources. We consider two sources: IS-IS Link State Packets (LSPs) and Syslog messages. Our first results on the Abilene and GÉANT networks show that the inference methods behave differently and that using IS-IS LSPs provides more accurate results than using Syslog messages.