Paris (France), Octobre 29-30 2018
Paper Submission: | June 10, 2018 Extended: July 1st, 2018 |
Authors Notification: | July 27th, 2018 |
Camera ready: | August 20th, 2018 |
Conference: | October 29-30, 2018 |
General Chairs:
Paolo Ballarini, Centrale Supelec, Université Paris Saclay
Benoît Barbot, Université Paris Est Créteil
Hind Castel, Télécom SudParis, Institut mines télécom
Program Co-Chairs
Rena Bakhshi, Netherlands eScience Center
Anne Remke, WWU Münster
Elvio Gilberto Amparore, Università degli studi di Torino
Urtzi Ayesta, CNRS IRIT & LAAS
Paolo Ballarini, CentraleSupelec
Simonetta Balsamo, Dipartimento di Informatica Università Ca' Foscari di Venezia
Benoît Barbot, Université Paris-Est Créteil
Marco Beccuti, Università degli studi di Torino
Marco Bernardo, University of Urbino
Marco Biagi, University of Florence
Olivier Brun, LAAS-CNRS
Laura Carnevali, University of Florence
Hind Castel, Telecom SudParis
Tadeusz Czachórski, Institute of Theoretical and Applied Informatics, Polish Academy of Sciences, Gliwice
Dieter Fiems, Ghent University
Jean-Michel Fourneau, University of Versailles
Boudewijn Haverkort, University of Twente
Stephen Gilmore, The University of Edinburgh
András Horváth, University of Turin
Gábor Horváth, Budapest University of Technology and Economics
Emmanuel Hyon, Sorbonne University
Alain Jean-Marie, INRIA
William Knottenbelt, Imperial College London
Samuel Kounev, University of Wuerzburg
Lasse Leskelä, Aalto University
Catalina Llado, University of the Balearic Islands
Andrea Marin, University of venice
Lynda Mokdad, Université Paris-Est Créteil
Marco Paolieri, University of Southern California
Nihal Pekergin, Universite Paris-Est Créteil
Agapios Platis, University of the Aegean
Philipp Reinecke, Cardiff University
Markus Siegle, Uni Bw Munich
Miklos Telek, Budapest University of Technology and Economics
Nigel Thomas, Newcastle University
Petr Tuma, Charles University
Enrico Vicario, Florence University
Jean-Marc Vincent, Université Grenoble-Alpes
Joris Walraevens, Ghent University
Qiushi Wang, Nanyang Technological University
Katinka Wolter, Frei Universität Berlin
Huaming Wu, Tianjin University
EPEW 2017 (Berlin, Germany)
EPEW 2016 (Chios, Greece)
EPEW 2015 (Madrid, Spain)
EPEW 2014 (Florence, Italy)
EPEW 2013 (Venice, Italy)
EPEW 2012 (Munich, Germany)
EPEW 2011 (Borrowdale, UK)
EPEW 2010 (Bertinoro, Italy)
EPEW 2009 (Imperial College, UK)
EPEW 2008 (Palma de Mallorca, Spain)
EPEW 2007 (Berlin, Germany)
EPEW 2006 (Budapest, Hungary)
EPEW 2005 (Versailles, France)
EPEW 2004 (Toledo, Spain)
Proceedings: Springer LNCS
The European Performance Engineering Workshop is an annual event that aims to gather academic and industrial researchers working on all aspects of performance engineering.
We invite original papers related to the following areas:
Submissions must be original and should not have been published previously or be under consideration for publication. Accepted papers will be published in the conference proceedings, in the Lecture Notes in Computer Science series, by Springer-Verlag. Authors will be required to sign a copyright release.
Extended versions of the best papers will be considered for possible fast-track publication in the ACM Transactions on Modeling and Performance Evaluation of Computing Systems (TOMPECS).
Papers must not exceed 15 pages in camera-ready form, including figures and references. All papers must use the Springer LNCS style, available HERE. All submissions must be in English.
All submissions are handled by our Easychair website. Please click HERE to access it.
Abstract:
Performance evaluation targeting Quality of Experience
Nowadays, Quality of Experience (QoE) is taking the lead when evaluating applications or services operating on the Internet and centered on audio or video content (and composing most of today's and tomorrow's Internet traffic). Our performance evaluation tools lead with "lower level" metrics, such as throughputs, response times, utilization rates, etc. QoE is associated with a higher level view, depending in a complex way on those low level parameters. In the presentation, we will propose an approach allowing to use classic performance evaluation tools (queuing models, SPNs, Markov models,...) and still evaluate the QoE directly. It is based on the possibility of developing an approach for measuring QoE that can be seen as analytical, in some sense. We will illustrate this idea with the author's own approach called PSQA, for real time measuring of the QoE, and based on Machine Learning tools.
Abstract:
Mean field models for (large-scale) load balancing systems
This talk focuses on the behavior of load balancing systems and is composed of three parts. In the first part we revisit a classic mean field result on load balancing in large distributed systems. More specifically, we focus on the celebrated power-of-two choices paradigm and its mean field limit. We subsequently introduce a theorem for the class of density dependent population processes established in the 1970s by Kurtz and discuss some of the technical issues involved to extend this result to the stationary regime. We end the
first part by illustrating the accuracy of the mean field limit using simulation.
In the second part we introduce the refined mean field approximation, which is a tech- nique to compute a 1/N correction term to improve the accuracy of classic mean field limits. This technique can be used to more accurately approximate the performance of small systems, e.g., consisting of N = 10 servers, and can be applied to any density depen- dent population processes with limited effort. We focus on the different computational steps involved to compute this correction term and illustrate its accuracy on various numerical
examples.
In the final part of the talk we discuss some recent results on load balancing schemes
that select servers based on workload information (as opposed to queue length information). Such systems are motivated by load balancing systems that use late binding or redundancy. We present explicit results for the workload and response time distribution when the job sizes follow an exponential distribution and indicate how to compute these distributions for non-exponential jobs sizes.
This talk is based on joint work with Nicolas Gast (Inria) and Tim Hellemans (Univer- sity of Antwerp).
The proceeding is available here.
Time | Monday Oct. 29th | Tuesday Oct. 30th |
---|---|---|
09:00 | Registration, Opening and Coffee | Opening and Coffee |
10:00 | EPEW Keynote I : (chair : Isi Mitrani) Mean field models for (large-scale) load balancing systems Prof. Benny Van Houdt, University of Antwerpen, Belgium |
EPEW Keynote II : (chair : Peter Harrison) Performance evaluation targeting Quality of Experience Prof. Gerardo Rubino, INRIA, Rennes |
11:00 | Coffee break | Coffee break |
11:30 |
Advances on Stochastic Analysis : Benoît Barbot Deriving Symbolic Ordinary Differential Equations from Stochastic Symmetric Nets without unfolding Information Flow Security for Stochastic Models Towards Probabilistic Modeling and Analysis of Real-Time Systems |
Product-Forms, Steady-States Solutions : Olivier Brun
Product-form queueing networks with batches
Mean Value Analysis of closed G-networks with signals Extending the steady state analysis of hierarchical semi-Markov processes with parallel regions |
13:00 | Lunch break |
Lunch break |
14:30 |
Stochastic Modeling : Giuliana Franceschinis
An Ontology Framework for Generating Discrete-Event Stochastic Models Formal Parameter Synthesis for Energy-Utility-Optimal Fault Tolerance Modelling Smart Buildings using Fault Maintenance Trees To What Extents does Performance Awareness Support Developers in Fixing Performance Bugs? |
Network Performance Modeling : Nigel Thomas Modeling the Effect of Parallel Execution on Multi-site Computation Offloading in Mobile Cloud Computing
A mixed strategy for a competitive game in Delay Tolerant Network An OpenFlow Controller Performance Evaluation Tool |
16:00 | Coffee break | |
16:30 | Coffee break |
Case Studies : Laura Carnevali
Performance Model of Apache Cassandra under Heterogeneous Workload using the Quantitative Verification Approach Performance Impact of Misbehaving Voters |
17:00 |
Fluid Performance Modeling : Josu Doncel On the degradation of distributed graph databases with eventual consistency Second Order Fluid Performance Evaluation Models for Interactive 3D Multimedia Streaming |
|
17:30 | Closing | |
18:00 | ||
19:00 | Dinner |
The conference will be held at the Télécom ParisTech, in Paris, France
The conference dinner will be held at Les noces de jeannette.
The conference takes place in the centre of Paris. Many hotels of various prices can be found close to the conference venue. You can find some hotel on this website here and there.
Paolo Ballarini: paolo[dot]ballarini[at]ecp[dot]fr
Benoît Barbot: benoit[dot]barbot[at]lacl[dot]fr
Hind Castel: hind[dot]castel[at]telecom-sudparis[dot]eu