Context

Data centers are computing infrastructures that host most of the services available on the internet. As data centers may include thousands of servers, both their energy consumption and their carbon footprint are significant. This has led many research projects to focus on optimizing these two objectives. This PhD takes place in the context of centralized (Cloud) and decentralized (Edge, Fog) infrastructures.

Various ideas emerge from research / R&D to reduce the impact of computing infrastructures, and these ideas must be evaluated. Doing such evaluations on real infrastructures is unfeasible, as this would induce a tremendous time and energy cost. Consequently, most evaluations are instead done in simulation, where models of applications are executed on models of computing resources. In this PhD, we will focus on the SimGrid-based simulator Batsim.

Batsim has been designed with batch applications in mind. Batch applications first read input data, then issue some computations, then finally write output data and terminate. Batch applications are omnipresent in high performance computing (the context from which Batsim originated), but Cloud and Edge infrastructure mostly host service applications instead. Service applications have a long (infinite) lifecycle and adapt their shape and behavior dynamically depending on the load. As we write these lines, Batsim does not easily enable to model service applications.

Objective

The main objective of this PhD is to simulate service applications accurately. Technically, new service application models will be implemented in Batsim and these models will be evaluated. The main tasks of this PhD are the following.

• Conduct a bibliographic survey about the service application models available in service-oriented (Cloud/Edge/Fog/…) simulators.
• Select service application models from the literature, implement them in Batsim, and evaluate their accuracy. Evaluation will be done by comparing simulated executions against real executions of real service applications. Real executions will be conducted on the Grid'5000 testbed.
• Design and implement service placement strategies (algorithms) that optimize the energy consumption or the carbon footprint of a distributed Edge infrastructure. These strategies will be evaluated in simulation.

Expected skills and profile

• Required: Currently in a master’s in computer science.
• Required: Fluent French or English.
• Strongly recommended: A taste for experimental approaches, C, C++ or Python programming.
• Recommended: A taste for contributing to open source projects.
• Appreciated: Background in optimization, performance evaluation and modeling.

At the end of the PhD, the student will have acquired the following skills: collaborative development of software, expertise in cloud and edge systems, planning of long term projects, scientific writing.

Practical details

The PhD will take place at IRIT, the largest computer science research institute in Toulouse, France. Our team SEPIA works on resource management on various distributed systems (cloud datacenters, HPC centers, edge architectures, IoT…) and is especially interested in ecological transition, notably by reducing energy consumption and CO2 emissions, by using renewable energy.

It will be supervised by Millian Poquet and Patricia Stolf in a convivial atmosphere :).

This PhD will be funded by the CareCloud project of the PEPR Cloud collaborative project. This project involves many academic partners in France, the PhD will therefore be done in a national collaboration context. It is expected to have three annual meetings in other towns in France during which English will be used. The monthly gross salary is slighly more than 2050 €.

The student will have the opportunity to teach in English or in French if she or he wishes.

You can send us your application (cover letter + resume + transcript of records for the full bachelor and current master) by email to millian.poquet@irit.fr and patricia.stolf@irit.fr