[ecoop-info] Post-doc position at Inria (Stack team @ Nantes)

Thomas Ledoux Thomas.Ledoux at imt-atlantique.fr
Thu Mar 22 13:33:33 CET 2018


The Stack team of Inria (http://stack.inria.fr) is looking for a 
postdoctoral fellow in Cloud autonomic computing in collaboration with 
Inria Ctrl-A team (https://team.inria.fr/ctrl-a).

--- Subject
Autonomic controllers for the self-adaptation of micro-services in 
Fog/Edge and Cloud computing

--- Context
Cloud systems (Cloud and now Fog/Edge computing) are becoming so complex 
that they cannot be handled manually in an efficient way. This concerns 
not only the base configuration and monitoring activities, but also the 
way Cloud systems should behave at runtime in order to guarantee certain 
QoS (/Quality of Service/) levels and expected SLA (/Service Level 
Agreement/) contracts. Autonomic Computing [1] has been largely adopted 
to tackle that kind of dynamic environments. In order to achieve 
well-tuned Autonomic controllers (ACs), we need for design methods for 
well-behaved adaptation control.
The topic of this proposal is to work on design methods for such 
systems, following approaches based on behavioral models and control 
techniques. The goal is to obtain ACs that ensure safety of the system 
executions and correct fulfillment of the SLA objectives.
In the general context of Autonomic Computing, explicitly designed 
/feedback loops/ are performing the management of adaptations at 
hardware and software levels. To go beyond the difficult and error-prone 
hand writing of management rules, some approaches explored controller 
design methods based on models of the dynamics of the platform [2, 4]. 
Another way to address the issue of achieving ACs is to use declarative 
programming describing what computation should be performed and not how 
to compute it. In a recent work [3], we complied with SLA contracts and 
Cloud internal constraints (e.g., maximum capacity of given resources) 
and we designed an AC -- based on a constraint solver -- that can find 
the optimal configurations based on these objectives. However, these 
/constraints-based loops /reason on relations between variables at a 
given moment and do not take into account feedback values, history of 
past values or decisions, dynamics of the system, predictive decision or 
apprenticeship.

--- Objectives
The topic of this post-doctorate position is to work on defining more 
general and powerful Autonomic controllers, by combining several 
modeling and decision approaches, in a framework of coordination of 
multiple autonomic loops [4, 5].
The motivation stems from the fact that complex Cloud and Fog systems 
present a variety of regulation problems of different natures, to be 
solved together, and for which different formalization and solving 
techniques are required (e.g., how to address Cloud elasticity 
oscillations, how to address edge node failure).
The person recruited will be in charge of considering combinations of 
controllers involving constraints-based approaches, behavioral models 
such as discrete controllers or continuous models from Control Theory. 
He will define methods that enable to ensure safety of the combined 
complementary control loops, by proposing /coordination patterns/. For 
example, constraint-solving can be applied on variables, some if which 
are computed in a control feedback loop.

-- Main activities
The work will be organized around several axes:

  * identification, analysis and description of important decision and
    control problems in Cloud and Fog/Edge computing, in order to focus
    the research work on relevant applicative problems;
  * design of generic methods for their decomposition into complementary
    sub-problems, formalized in the appropriate models, among the
    approaches mentioned above;
  * composition of the sub-controllers by coordination patterns insuring
    the absence of interferences between the loops;
  * validation of the proposed methods in a simulator of distributed
    applications in heterogeneous distributed environments (such as
    SimGrid). The use case will be a medical emergency scenario in the
    context of Fog computing. The application will be based on the
    micro-services paradigm.

--- Keywords
Cloud infrastructures, autonomic computing, control techniques, 
constraint solver, micro-services

--- References
[1] J. O. Kephart and D. M. Chess, “The vision of autonomic computing”, 
Computer, vol. 36, no. 1, pp. 41–50, Jan 2003.
[2] Nicolas Berthier, Eric Rutten, Noël de Palma, Soguy Mak-Karé Gueye. 
Designing Autonomic Management Systems by using Reactive Control 
Techniques. IEEE Transactions on Software Engineering, TSE. Volume: 42, 
Issue: 7, 1 July 2016, Pages 640 - 657.
[3] Jonathan Lejeune, Frederico Alvares and Thomas Ledoux. Towards a 
generic autonomic model to manage Cloud Services. In the 7th 
International Conference on Cloud Computing and Services Science (CLOSER 
2017), Porto, Portugal, 24-26  April 2017 (best paper)
[4] Soguy Mak-Karé Gueye, Noël de Palma, Eric Rutten, Alain Tchana, 
Nicolas Berthier. Coordinating self-sizing and self-repair Managers for 
Multi-tier Systems. journal of Future Generation Computer Systems Volume 
35, June 2014, pp. 14–26.
[5] Frederico Alvares de Oliveira Jr., Remi Sharrock and Thomas Ledoux. 
Synchronization of Multiple Autonomic Control Loops: Application to 
Cloud Computing. In the 14th International Conference on Coordination 
Models and Languages (Coordination 2012), Stockholm, Sweden, June 14-15, 
2012.

--- Advisors
The recruited person will be under the supervision of Thomas Ledoux 
(Stack <http://stack.inria.fr>@ Nantes) for /constraints-based loops/ 
and in connection with Eric Rutten (Ctrl-A 
<https://team.inria.fr/ctrl-a> @ Grenoble)  for/feedback loops/ .
Thomas Ledoux, Stack team (Thomas.Ledoux at inria.fr), IMT Atlantique, 
campus Nantes,  4, rue Alfred Kastler - BP 20722 - 44307 Nantes Cedex 3
Eric Rutten, Ctrl-A team (Eric.Rutten at inria.fr), Inria Grenoble - 
Rhône-Alpes,  Antenne GIANT, Batiment 50C Minatec, 17 rue des Martyrs 
38054 Grenoble Cedex

--- Candidate profile and skills
Technical skills and level required: the candidate must hold a PhD in 
Computer Sciences with advanced knowledge on Cloud architectures and 
Control techniques.
Languages: strong oral and written communication skills in English, 
including the ability to publish the results of scientific research in 
scientific journals
Relational skills: curiosity, autonomy and social capabilities
Other value appreciated: good programming knowledge

-- Conditions
Location: IMT Atlantique, campus de Nantes (+ some stays in Grenoble are 
expected)
Duration: 12 months
Remuneration: Monthly gross salary amounting to 2653 euros
Expected starting date: October/November 2018 (the candidate must have 
defended his thesis before taking office)

-- Application
*Selection process: due to funding conditions (**Inria post-doc 
competition), the candidate must submit a very good application
**Deadline for application: **April 2, 2018*
Interested candidates should send the following documents to Thomas 
Ledoux <Thomas.Ledoux at inria.fr> listed below:
a CV, the list of publications accompanied by the two publications 
judged the most representative, a letter of motivation, one or more 
letters of recommendation, a perspective of professional insertion after 
the post-doctorate.

-- 
IMT Atlantique <http://www.imt-atlantique.fr> 	
*Thomas LEDOUX*
Enseignant Chercheur (équipe Stack - DAPI)
Responsable pédagogique FIL - 02 51 85 82 19
La Chantrerie 4 rue Alfred Kastler BP 20722
44307 Nantes Cedex 3
Site web IMT Atlantique <http://www.imt-atlantique.fr>Twitter IMT 
Atlantique <https://twitter.com/IMTAtlantique>Facebook IMT Atlantique 
<https://www.facebook.com/IMTAtlantique>Google+ IMT Atlantique 
<https://plus.google.com/109356934816329928502>LinkedIn IMT Atlantique 
<https://www.linkedin.com/company/imt-france>Blog IMT Atlantique 
<http://blogrecherche.wp.imt.fr>
Une école de l'IMT <https://www.imt.fr>



More information about the ecoop-info mailing list