Jad Darrous

Jad Darrous

Ph.D. in Large-scale Data Management Systems.

About me

[IMPORTANT]

Successfully defended my Ph.D. on Tuesday, 17 December at ENS de Lyon. The thesis is available here while the presentation slides are here.

Research interests

  • Data management
  • Erasure coding
  • MapReduce framework
  • Distributed (storage) systems
  • Fog and Edge clouds
  • Geo-distributed clouds

Research team

I’m a member of Avalon and Stack teams of inria research center that are hosted in LIP and LS2N laboratories, respectively. I’m registered at the university (i.e., école) ENS de Lyon for the Ph.D. degree. During my Ph.D., I’m working closely with my two advisors: Shadi Ibrahim and Christian Perez.

Ph.D. Topic

I’m working on topics related to large-scale Data Management in geographically distributed clouds and Edge environments.

The first part of my Ph.D. was dedicated to improve service provisioning in geo-distributed clouds and Fog/Edge environments. We worked on network-aware data placement and retrieval of Virtual Machine Images (VMIs) and container images in geo-distributed clouds and and Edge environments (for more details, our work led to two publications here and here).

Currently, I’m focusing on reducing the storage cost of data-intensive clusters without scarifying data availability. More specifically, we study the opportunities and the challenges in integrating Erasure Coding (EC) for online data analytics. As a first step, through an in-depth performance evaluation study we evaluate how analytic workloads, data persistency, failures, the back-end storage devices, and the network configuration impact the performance for MapReduce jobs running under erasure coding and replication (this work led to a publication here). Moreover, a poster about data processing in Fog can be found here.

Briefly, the subject of my thesis is about scalable and efficient data management for building and running data-intensive services.

Research Project

I’m part of DISCOVERY Initiative, a research project that explores the methodologies and algorithms needed to deploy and operate massively distributed cloud infrastructure as Fog/Edge environments. Developed prototypes and tools are ported to OpenStack, an open source de-facto IaaS manager.

Papers reviews

I reviewed papers as subreviewer for ICPP 2019, SmartData 2019, HPBDC 2019, Cluster 2017, CCGRID 2017 and CloudCom 2017.