Search

International audience; Une cartographie plus complète de la topologie d'internet permet de mieux comprendre des propriétés importantes d'internet, comme sa résilience et sa performance. Nous présentons deux contributions pour fournir une topologie plus complète à la communauté : (1) Zeph, un algorithme d'apprentissage par renforcement qui alloue les préfixes à sonder pour chaque point de mesures afin d'optimiser la découverte de la topologie IP d'internet. (2) Iris, une plateforme de mesure open-source capable de conduire des mesures de topologie exhaustives sur l'ensemble d'internet.

International audience; EdgeNet is a software-only distributed systems testbed in the family of PlanetLab. It is realized as a multitenant global Kubernetes cluster running on volunteered virtual machines. In this paper, we describe our modifications to Kubernetes which enable trusted execution with accountability on untrusted, volunteered hardware, user control of the placement of Kubernetes pods, and control of a global cluster from a single central head node.

International audience; EdgeNet is a public Kubernetes cluster dedicated to network and distributed systems research, supporting experiments that are deployed concurrently by independent groups. Its nodes are hosted by multiple institutions around the world. It represents a departure from the classic Kubernetes model, where the nodes that are available to a single tenant reside in a small number of well-interconnected data centers. The free open-source EdgeNet code extends Kubernetes to the edge, making three key contributions: multi-tenancy, geographical deployments, and single-command node installation. We show that establishing a public Kubernetes cluster over the internet, with multiple tenants and multiple hosting providers is viable. Preliminary results also indicate that the EdgeNet testbed that we run provides a satisfactory environment to run a variety of experiments with minimal network overhead.

International audience; IP geolocation has myriad applications. While a body of prior research has investigated the accuracy of geolocation databases, we take a first look at their stability. Using a large collection of snapshots from a popular geolocation database, we examine the longitudinal evolution of its location mappings and address coverage. Across different classes of IP addresses, we find that significant differences can exist even between two successive weekly snapshots - a previously underappreciated source of potential error. To assess the sensitivity of research results to the geo database instance, we examine a prior study [1] that used geolocation. Using their data and methodology, we generate results for each database instance available during their measurement period, i.e., the hypothetical results had the authors used a different snapshot. We show that the median distance of addresses considered shifted over 100km from ground truth and the coverage differed by 30% - potentially impacting the conclusions of this prior study. Based on our findings, we recommend best practices when using geolocation databases for network research to encourage reproducibility and soundness.

Alias resolution techniques (e.g., Midar) associate, mostly through active measurement, a set of IP addresses as belonging to a common router. These techniques rely on distinct router features that can serve as a signature. Their applicability is affected by router support of the features and the robustness of the signature. This paper presents a new alias resolution tool called Limited Ltd. that exploits ICMP rate limiting, a feature that is increasingly supported by modern routers that has not previously been used for alias resolution. It sends ICMP probes toward target interfaces in order to trigger rate limiting, extracting features from the probe reply loss traces. It uses a machine learning classifier to designate pairs of interfaces as aliases. We describe the details of the algorithm used by Limited Ltd. and illustrate its feasibility and accuracy. Limited Ltd. not only is the first tool that can perform alias resolution on IPv6 routers that do not generate monotonically increasing fragmentation IDs (e.g., Juniper routers) but it also complements the state-of-the-art techniques for IPv4 alias resolution. All of our code and the collected dataset are publicly available., Comment: Preprint to appear in Proceedings of Passive and Active Measurement (PAM 2020) Conference, Eugene, OR, March 2020

International audience; In this paper, we set up measurements and make performance and geographic analysis of YouTube CDN video platform. We use large distributed testbeds, like PlatnetLab and EdgeNet, to grasp the heterogeneity of client situations. Those facilities can work as real clients without any simulation. From these infrastructures, we generate numerous requests to YouTube video servers. Using a reduced initial set of nodes in different geographic location, we continuously measure information related to YouTube homepage websites and video servers, and calculate the latency from clients to cache servers. We also look at the geographical location of YouTube servers. This enables a better understanding of cache mapping strategy and draws the map of the system. Our first result focus on distance between users and data centers before studying dynamic aspect of the system. The information we collect can be of interest to e.g. ISP network operators who need to improve their network architecture to minimize costs and optimize quality for the user.

Since its introduction in 2006-2007, Paris Traceroute and its Multipath Detection Algorithm (MDA) have been used to conduct well over a billion IP level multipath route traces from platforms such as M-Lab. Unfortunately, the MDA requires a large number of packets in order to trace an entire topology of load balanced paths between a source and a destination, which makes it undesirable for platforms that otherwise deploy Paris Traceroute, such as RIPE Atlas. In this paper we present a major update to the Paris Traceroute tool. Our contributions are: (1) MDA-Lite, an alternative to the MDA that significantly cuts overhead while maintaining a low failure probability; (2) Fakeroute, a simulator that enables validation of a multipath route tracing tool's adherence to its claimed failure probability bounds; (3) multilevel multipath route tracing, with, for the first time, a Traceroute tool that provides a router-level view of multipath routes; and (4) surveys at both the IP and router levels of multipath routing in the Internet, showing, among other things, that load balancing topologies have increased in size well beyond what has been previously reported as recently as 2016. The data and the software underlying these results are publicly available., Comment: Preprint. To appear in Proc. ACM Internet Measurement Conference 2018

International audience; To satisfy the need for ubiquitous connectivity of their customers, many enterprises (e.g., hotels, cafes and commercial centers) deploy EWLAN (Enterprise WLAN). In these networks, video streaming is both one of the most popular and challenging applications. To improve the overall users QoE in EWLAN, we propose to address simultaneously the resource bottlenecks that can occur on wireless links as well as the lack of overall bandwidth that can be experienced on the Internet access link. We propose a dynamic video Streaming Management based on software defined enterprIse wireLEss LANs (SMILE), an SDN framework which dynamically allocates the bandwidth among different access points based on their workloads and on the available bandwidth. We evaluate SMILE over several adaptive video streaming performance metrics: instability, rebuffering, and average video quality. The results demonstrate an improvement for these parameters with different adaptation logic's when SMILE is in use.

International audience; Dynamic Adaptive Streaming over HTTP (DASH) was introduced to enable high video quality streaming over HTTP. DASH depends on the adaptation logic at the client to choose which video bitrate to stream from the content server for each chunk. For clients receiving video over a cellular network, the cellular first hop tends to be the bandwidth bottleneck and can exhibit significant swings in available bandwidth. In this paper we develop and evaluate a Dynamic Adaptation for mobile Video Streaming (DAVS), a technique that can be used within DASH adaptation to handle the significant bandwidth variability experienced by cellular mobile clients. In our scheme the main innovation is that the client chooses a bitrate based on whether the playout buffer occupancy (BO) falls below or above a dynamic threshold. In addition, the scheme attempts to minimize bitrate switching by again delaying a change of video bitrate selection by a window of time-that is also dynamically determined. We evaluate the performance of DAVS over real traces collected from a mobile network operator. DAVS shows better performance over different video streaming metrics. Furthermore, It increases the QoE by a range 15%-55% compared to benchmark algorithms.

International audience; FORGE (http://ict-forge.eu/) was a European project that ran from 2013 to 2016 that developed the use of computer networking testbeds in e-learning. Through FORGE,teachers and students were given access to testbeds from the European Commission’s FIRE initiative, and were able to create and execute scientific experiments on these testbeds in ways that were suitable to the classroom or self-study. The e-learning tools developed by FORGE have been made available as open educational resources.We describe our work in FORGE, which consisted of making the PlanetLab Europe testbed (http://www.planet-lab.eu/) available to a MOOC on internet measurements, so that MOOC students could conduct lab exercises in a real-world environment. The MOOC is described here: https://www.fun-mooc.fr/courses/inria/41011/session01/aboutThose who are familiar with PlanetLab know that it offers SSH access to a command line environment in which one has root privileges on Linux virtual machines at servers around the world. This sort of access is appropriate for researchers engaging in long-term scientific experiments, but was ill adapted for students in this mooc for a number of reasons, among them: offering individual PlanetLab Europe accounts to 1800 students was not scalable and was not compatible with the privacy policy of the mooc platform; purely anonymous access would be incompatible with the PlanetLab security policies; the mooc students were a diverse lot and we could not count upon their knowledge of the Linux command line environment; there was a risk of flash crowd effects disrupting PlanetLab Europe’s nodes and servers.Our solution was to develop a dedicated client-server system for mooc measurements including ping, traceroute, and iperf (this latter requiring the coordination of iperf processes on two separate PlanetLab nodes at a time). The client component, run as part of the mooc environment, provided an individualized control panel which each student could use to request measurements and see their own results. The server system is accessed through a RESTful webservices interface. The server does not learn the identities of the students, but a chain of responsibility is maintained, in case of any disruptive measurements, as the mooc administrators can identify their own students on the basis of job numbers provided them by the PlanetLab Europe administrators. The server maintains a queue of measurement requests and feeds them to PlanetLab Europe nodes at a controlled rate. Some basic sanity checks are performed to reduce the potential for disruptive measurements. The server returns standard output, standard error, the process return code, and beginning and ending timestamps for the measurement. This client-server system is currently being extended to enable measurements in other courses and on other testbeds

In recent years, video traffic has been increasing due to the huge demands of video streaming services. Peer-to-peer (P2P) video streaming services (P2PTV) have attracted attention as a solution to decrease the server load because P2P communication can distribute the video data to peers. However, the random peer selection mechanisms of P2PTV conduce to the worldwide spread of P2PTV traffic, and thus it is a large issue to analyze the characteristics of P2PTV traffic. Several traffic measurements have been studied and revealed the characteristics of P2PTV traffic. However, these studies did not focus on users' behavior that has influence on traffic and also did not construct a P2PTV traffic model. In this paper, we analyzed the characteristics of P2PTV traffic that occur in watching a content to construct the P2PTV traffic model of each user. Moreover, we discussed how to model the P2PTV traffic. As a result, we can revealed several characteristics of P2PTV traffic such as the stable traffic and the bursty traffic. Furthermore, we extracted some statistical information and stated modeling the P2PTV traffic.