Limited-persistence distributed objects. Shared data is at the core of distributed applications, either due to the intrinsic nature of the application (e.g., social networks intended for sharing content) or to scale computation. Classically data persistence is a fundamental requirement: once data is changed, this change must be permanent unless and until it is overwritten. Unfortunately, guaranteeing persistence while ensuring consistency and data availability is a complex and expensive task and may even be impossible in failure-prone networks. While full persistence is fundamental for many critical applications (e.g., banking), limited data persistence is a key functionality of some distributed applications (for instance, the Snapchat social network attracts users based on the premise that user data is accessible only for a user-defined period and is deleted after that). Also some loss of information can be tolerated by real-time applications like video streaming to gain performance. Finally, there is a growing concern that keeping user-sensitive data for an unrestricted period may have adverse effects on users' privacy.
Motivated by these concerns, the objectives of the PhD thesis are to define objects with limited persistence, to investigate algorithmic techniques for implementing them and to gain a theoretical understanding of their implications on correctness, computability and complexity.
Laboratoire Bordelais de Recherche en Informatique (LaBRI). The project is connected to a wider Indo-French collaboration, manifested by the UMI ReLaX, and the student hired should expect to travel to Chennai, India, during his/her PhD.
Eligibility and formalities
Applicants should hold a Master's degree in computer science, at the latest at the time their contract starts. There is no nationality condition. To apply, please send the following documents at your earliest convenience to Alessia Milani.
Transcripts of undergraduate university grades,
Short text explaining the reasons of your interest for this proposal.
PhD position in distributed computing
at the University of Bordeaux/LaBRI