Doctorant F/H PhD Offer: Taming Experimentation for Distributed Systems from Testbed to Conduct Tools with Reproducible Guarantee

Le Centre Inria de l’Université Grenoble Alpes, regroupe un peu moins de 600 personnes réparties au sein de 23 équipes de recherche et 9 services support à la recherche.

Son effectif est distribué sur 3 campus à Grenoble, en lien étroit avec les laboratoires et les établissements de recherche et d'enseignement supérieur (Université Grenoble Alpes, CNRS, CEA, INRAE, …), mais aussi avec les acteurs économiques du territoire.

Présent dans les domaines du calcul et grands systèmes distribués, logiciels sûrs et systèmes embarqués, la modélisation de l’environnement à différentes échelles et la science des données et intelligence artificielle, le Centre Inria de l’Université Grenoble Alpes participe au meilleur niveau à la vie scientifique internationale par les résultats obtenus et les collaborations tant en Europe que dans le reste du monde.

Experimenting with distributed systems, particularly in High-Performance Computing (HPC) and hybrid HPC-Cloud environments, is notably tedious and complex when the entire software stack is taken into account. Numerous aspects must be managed, ranging from experiment design, software stack selection and preparation, resource provisioning, experiment development, debugging, and execution, to results retrieval and analysis. This inherent complexity significantly exacerbates the challenge of reproducing scientific results.

In recent work [1, 2], we demonstrated that by leveraging a functional approach to package complete software stacks [3], it is possible to obtain portable experiments with minimal modifications while maintaining software reproducibility guarantees. This approach allowed to switch between various testbeds, including containers, virtual machines, and bare-metal nodes. By simplifying the development and debugging phases, which are typically a considerable burden when studying complex distributed systems on HPC and Cloud platforms, this method also enables to consider various modifications and manipulation on the entire software stack using functional paradigms.

The objective of this PhD project is to build upon the work initiated in [1, 2] by incorporating additional elements involved in experimentation, such as experiment descriptions and testbed specificities. The goal is to further reduce the effort required to conduct robust experiments on complex distributed systems, thereby enhancing reproducibility and scalability.

This research will be conducted of the Exa-Atow/Numpex project, which addresses challenges associated with executing single- and multi-site HPC workflows and hybrid infrastructures.

References:

[1] Guilloteau, Quentin, et al. "Painless transposition of reproducible distributed environments with nixos compose." 2022 IEEE International Conference on Cluster Computing (CLUSTER). IEEE, 2022. Xiaohua Zhai et al., Scaling Vision Transformers, IEEE/CVF CVPR, 2022

[2] Guilloteau, Quentin. Control-based runtime management of HPC systems with support for reproducible experiments. Diss. Université Grenoble Alpes, 2023.

[3] Dolstra, Eelco. The purely functional software deployment model. Utrecht University, 2006.

We are seeking a talented and passionate student to join our high-performance computing
(HPC) research team to explore and propose methodologies and tools taming experimentation for distributed systems on actual and future testbeds.

The PhD candidate will focus on:

1. Formalizing from Experimentation Elements:
- Develop a formal framework to describe and abstract experimentation elements from all stack softwares, tools to experiment exection,
configuration testbed a configuration ensuring consistency and reproducibility.

2. Enhancing Testbed Tools with Nix/NixOS:
- Exploit Nix/NixOS’s declarative and functional paradigms to streamline configuration, deployment, and reconfiguration tasks.
- Extend system profiling capabilities to handle diverse target environments (e.g., machines, VMs, containers).

3. Abstracting Platform Diversity:
- Build on the NixOS-Compose tool to simplify transitions across experimental setups (e.g., containers, VMs, Grid5000).
- Integrate tools for describing and managing large-scale testbeds, creating innovative solutions like a "Lite-G5K" with Nix/NixOS.

4. Guaranteeing Reproducibility in Experimentation:
- Address the combinatorial explosion of configurations by designing reproducibility-focused methodologies.
- Develop formal models and tools to systematically validate experimentation workflows and outcomes.

Mandatory skills (the candidate must have):

Strong knowledge of computer system (system adminstration, network, distributed system)
Strong programming knowledge of Functional and Imperatives languages, and System Tools
Knowledge in formal methods

Languages: proficiency in written English and fluency in spoken English.

Relational skills: the candidate will work in a research team, where regular meetings will be set up. The candidate has to be able to present the progress of their work in a clear and detailed manner.

Other values appreciated: Open-mindedness, strong integration skills, and team spirit.

• Restauration subventionnée
• Transports publics remboursés partiellement
• Congés: 7 semaines de congés annuels + 10 jours de RTT (base temps plein) + possibilité d'autorisations d'absence exceptionnelle (ex : enfants malades, déménagement)
• Possibilité de télétravail (après 6 mois d'ancienneté) et aménagement du temps de travail
• Équipements professionnels à disposition (visioconférence, prêts de matériels informatiques, etc.)
• Prestations sociales, culturelles et sportives (Association de gestion des œuvres sociales d'Inria)
• Accès à la formation professionnelle
• Participation employeur mutuelle santé (sous conditions)

2200 euros brut /mois