SEMINARS

MonthWeekDay
December 2020
Monday Tuesday Wednesday Thursday Friday Saturday Sunday
November 30, 2020 December 1, 2020 December 2, 2020 December 3, 2020 December 4, 2020 December 5, 2020 December 6, 2020
December 7, 2020 December 8, 2020 December 9, 2020 December 10, 2020 December 11, 2020 December 12, 2020 December 13, 2020
December 14, 2020 December 15, 2020 December 16, 2020 December 17, 2020 December 18, 2020 December 19, 2020 December 20, 2020
December 21, 2020 December 22, 2020 December 23, 2020 December 24, 2020 December 25, 2020 December 26, 2020 December 27, 2020
December 28, 2020 December 29, 2020 December 30, 2020 December 31, 2020 January 1, 2021 January 2, 2021 January 3, 2021
  • July 2, 2020 @ -- Seminar: Salah Zrigui

    Title: Classification of energy profiles of HPC tasks

    Abstract: Salah will present how he analyzed, characterized and tried to classify energy consumption traces (time series) of HPC codes running on one of the GRICAD cluster.

  • October 23, 2020 @ Bâtiment IMAG (amphitheater) -- PhD defense of Baptiste Jonglez: End-to-end mechanisms to improve latency in communication networks

    It is my pleasure to invite you to my PhD defense whose subject is " End-to-end mechanisms to improve latency in communication networks". The defense will take place on October 23 at 2 pm, with two possible ways to attend:

    The defence will take place in English. The jury is composed of:

    • M. Martin Heusse, Professor, Grenoble INP, PhD supervisor
    • M. Bruno Gaujal, Directeur de recherche, Inria, PhD co-supervisor
    • M. André-Luc Beylot, Professor, INP Toulouse - ENSEEIHT, Referee
    • M. Guillaume Urvoy-Keller, Professor, Université Côte d'Azur, Referee
    • Mme Isabelle Guérin Lassous, Professor, Université Lyon 1, Examiner
    • Mme Anna Brunström, Professor, Karlstads Universitet (Sweden), Examiner

    Here is the thesis abstract:

    The network technologies that underpin the Internet have evolved significantly over the last decades, but one aspect of network performance has remained relatively unchanged: latency. In 25 years, the typical capacity or "bandwidth" of transmission technologies has increased by 5 orders of magnitude, while latency has barely improved by an order of magnitude. Indeed, there are hard limits on latency, such as the propagation delay which remains ultimately bounded by the speed of light. This diverging evolution between capacity and latency is having a profound impact on protocol design and performance, especially in the area of transport protocols. It indirectly caused the Bufferbloat problem, whereby router buffers are persistently full, increasing latency even more. In addition, the requirements of end-users have changed, and they expect applications to be much more reactive. As a result, new techniques are needed to reduce the latency experienced by end-hosts. This thesis aims at reducing the experienced latency by using end-to-end mechanisms, as opposed to "infrastructure" mechanisms. Two end-to-end mechanisms are proposed. The first is to multiplex several messages or data flows into a single persistent connection. This allows better measurements of network conditions (latency, packet loss); this, in turn, enables better adaptation such as faster retransmission. I applied this technique to DNS messages, where I show that it significantly improves end-to-end latency in case of packet loss. However, depending on the transport protocol used, messages can suffer from Head-of-Line blocking: this problem can be solved by using QUIC or SCTP instead of TCP. The second proposed mechanism is to exploit multiple network paths (such as Wi-Fi, wired Ethernet, 4G). The idea is to use low-latency paths for latency-sensitive network traffic, while bulk traffic can still exploit the aggregated capacity of all paths. This idea was partially realized by Multipath TCP, but it lacks support for multiplexing. Adding multiplexing allows data flows to cooperate and ensures that the scheduler has better visibility on the needs of individual data flows. This effectively amounts to a scheduling problem that was identified only very recently in the literature as "stream-aware multipath scheduling". My first contribution is to model this scheduling problem. As a second contribution, I proposed a new stream-aware multipath scheduler, SRPT-ECF, that improves the performance of small flows without impacting larger flows. This scheduler could be implemented as part of a MPQUIC (Multipath QUIC) implementation. More generally, these results open new opportunities for cooperation between flows, with applications such as improving WAN aggregation.

  • November 4, 2020 @ Bâtiment IMAG (amphitheater) -- PhD defense of Bruno Donassolo: IoT Orchestration in the Fog

    I am pleased to invite you to my thesis defense entitled: « IoT Orchestration in the Fog ». This thesis has been conducted in the context of the <I/O lab>, a joint lab between Inria and Orange Labs. The defense will be in English and will take place in the auditorium of the IMAG building (700 Avenue Centrale, 38401 Saint-Martin-d'Hères).

    Video-conference link: https://meet.univ-grenoble-alpes.fr/b/bru-m9j-mfv

    You are also invited to share a “pot” following the defense (if the sanitary conditions allow)

    The members of jury are:

    • Mme E. Veronica BELMEGA, Maître de conférences, Université CY Cergy Paris, Reviewer
    • M. Adrien LEBRE, Professeur des Universités, IMT Atlantique, Reviewer
    • M. Frédéric DEPREZ, Directeur de recherche, INRIA Grenoble Rhône-Alpes, Examiner
    • Mme Nathalie MITTON, Directrice de recherche, Inria Lille-Nord Europe, Examiner
    • M. Ola ANGELSMARK, Ingénieur de recherche, Ericsson Research, Examiner
    • M. Arnaud LEGRAND, Directeur de recherche, CNRS, Supervisor
    • M. Panayotis MERTIKOPOULOS, Chargé de recherche, CNRS, Co-supervisor
    • Mme Ilhem FAJJARI, Ingénieur de recherche, Orange Labs, Co-supervisor

    Abstract: Internet of Things (IoT) continues its evolution, causing a drastically growth of traffic and processing demands. Consequently, 5G players are urged to rethink their infrastructures. In this context, Fog computing bridges the gap between Cloud and edge devices, providing nearby devices with analytics and data storage capabilities, increasing considerably the capacity of the infrastructure. However, the Fog raises several challenges which decelerate its adoption. Among them, the orchestration is crucial, handling the life-cycle management of IoT applications. In this thesis, we are mainly interested in: i) the provisioning problem, i.e., placing multi-component IoT applications on the heterogeneous Fog infrastructure; and ii) the reconfiguration problem, i.e., how to dynamically adapt the placement of applications, depending on application needs and evolution of resource usage. To perform the orchestration studies, we first propose FITOR, an orchestration system for IoT applications in the Fog environment. This solution addresses the lack of practical Fog solutions, creating a realistic environment on which we can evaluate the orchestration proposals. We study the Fog service provisioning issue in this practical environment. In this regard, we propose two novel strategies, O-FSP and GO-FSP, which optimize the placement of IoT application components while coping with their strict performance requirements. To do so, we first propose an Integer Linear Programming formulation for the IoT application provisioning problem. Based on extensive experiments, the results obtained show that the proposed strategies are able to decrease the provisioning cost while meeting the application requirements. Finally, we tackle the reconfiguration problem, proposing and evaluating a series of reconfiguration algorithms, based on both online scheduling and online learning approaches. Through an extensive set of experiments, we demonstrate that the performance strongly depends on the quality and availability of information from Fog infrastructure and IoT applications. In addition, we show that a reactive and greedy strategy can overcome the performance of state-of-the-art online learning algorithms, as long as the strategy has access to a little extra information.

Comments are closed.