WP1: Fault-tolerant High-Performance Storage

Leader: Thomas Ropars (ERODS), François Tessier (KerData)
Contributors: ERODS (UGA), KerData (Inria), TeraLab (IMT), DRIM (CNRS), Progress (UB)
Objectives: Take advantage of new storage technologies (NVRAM, CXL) to improve the performance and reliability of cloud storage solutions.

WP2: Data Storage and Processing on the Computing Continuum

Leader: Alexandru Costan (KerData)
Contributors: KerData (Inria), Myriads (Université de Rennes), STACK (IMT), TeraLab (IMT)
Objectives: WP2 aims to address the challenges of managing and storing data from geographically distributed applications deployed across the digital continuum. One of these challenges lies in geographical distribution, which means that parts of these applications will be run in different areas of the world. (e.g. remote sensors in the edge), subject to frequent network partitions, high node volatility and limited energy. Another problem stems from the heterogeneity of underlying infrastructures (i.e., edge, cloud, HPC) and available storage systems (i.e., volatile and non-volatile memories, distributed and parallel file systems). The objective of this WP is to propose a set of strategies to efficiently deal with problems related to network partitioning, fault tolerance and extreme heterogeneity. We will study the trade-offs of different solutions (e.g. relocation of datasets in case of network inaccessibility, application state backup, intelligent caches in disconnected mode, restarting computations on different nodes) from the point of view of performance and energy consumption.

Leader: DRIM (CNRS)
Contributors: DRIM (CNRS), Progress (UB), ERODS (UGA)
Objectives: Many online services used daily by millions of users require the collection, storage and analysis of large volumes of data of strategic interest or of a personal nature, requiring strong protection against unauthorized data access. Numerous software solutions exist, often based on cryptographic techniques, to enable secure storage on cloud infrastructures. Despite the emergence of hardware solutions (e.g., Intel SGX, AMD SEV, ARM Trustzone) that enable the construction of highly secure enclaves, many challenges remain open, including: (1) the storage of ultra-sensitive data; (2) the management of the heterogeneity of these secure enclaves; and (3) their use for the efficient handling of massive data. These challenges will be addressed within the framework of this WP.