Research
Background
My research background is in the field of distributed algorithms. More specifically, my expertise resides in solving important distributed problems (consensus, mutual exclusion, leader election, etc) in dynamic distributed systems (such as wireless networks or peer-to-peer systems), the properties of which do not fit the system model assumptions of traditional distributed algorithms literature. Solving distributed problems in such dynamic systems usually requires first modeling the dynamics of the system before designing an algorithm capable of progressing despite the system's inherent unstability.
My PhD thesis consisted in adapting the failure detector abstraction for dynamic systems in order to solve distributed problems.
Current research: the SkyData project
My current research revolves around the SkyData project. SkyData is an ANR project started in 2023.
A fundamental characteristic of our era is the deluge of Data. Data are everywhere, e.g., scientific applications, e-business, news, social networks. Each community develops its specialized data manager. Since Grid environments, data management in distributed environments is a quite common practice. To day Clouds provide many solutions to store data. A data manager can be defined through its functionalities that can be understood as services. Indeed, many services are required for data management, such as security, replication strategy, green data transfer, synchronization, data migration. Many ways to design those services exist and each data manager includes its own point of view to compose these services. Thus, when users need a specific data management, they can choose an appropriate data manager but with high dependency to this manager.
Usually the data management point of view is centered on applications rather than data, even when an autonomic solution is provided – for example in, authors deal with data placement for intensive web applications. Nonetheless, many approaches are relevant and efficient but domain-specific and do not provide any framework for the user to control its own data. Despite unquestionable benefits to users, Cloud computing raises several concerns about data management. Companies (Google, Apple, Huawei, Facebook and other social networks, Orange, Doctolib, etc.) have understood that whoever owns the data gets power and control over business. Now, information, which has been converted and stored in binary digital form, can be subject to security risks in terms of attacks, third-party use by companies, or loss due to lack of replication. This results in loss of control for users and companies.
Our proposal is to turn Data into Self-Managed Data. The SkyData project aims at breaking the existing rules and the way the data management has always been in place. We want to propose a new paradigm without any centralized control nor middleware. Imagine a world where the data are controlled by themselves. It is a real challenge to provide an autonomic behavior for the data. Nevertheless, with the progress of scientific research, we think it is now possible. We plan to build an algorithmic framework and companion tools enabling autonomy based on distributed algorithm knowledges and provide the genesis of a prototype with a new generation of data.
Within the project, my role is to (1) formalize distributed system models that capture the properties and dynamics of a SkyData environment, and (2) to provide algorithmic solutions to both existing and emerging distributed problems within a SkyData system.
Publications
My publications can be found on DBLP. Most are available as research reports on HAL.