Amadou DIARRA, R&D Engineer, University of Grenoble Alpes (UGA)

Aug 22 2017

How would you present the OCCIware project?

Cloud computing is becoming essential in many fields of today technologies. It provides a set of resources( that can be either software or hardware) that are remotely accessible as a service. A such system is very complex and involves lot of APIs and standards to access services. This presents many problems like interoperability, locking etc. For solving these problems, OCCI (Open Cloud Computing Interface) standard has been proposed by OGF (Open Grid Forum) to manage everything as service inside the cloud systems. But this standard is not formal and complete (it presents some defaults of conception).  So the  OCCIware project funded by PIA  has been proposed to address these shortcomings. 

What is your role in the project?

As member of OCCIware project, we work on quality of  service (QoS) aspects regarding the architecture of the technology, its runtime and the use cases..  More specifically, our goal is to guarantee a good level of performance in a cloud computing service using OCCI standard. 

What key innovation do you bring or help to develop?

The QoS techniques that we proprose for the OCCIware project are based on a state-of-the-art Performance debugging approach. Our QoS techniques aim at profiling and modelling a system in order to understand its performance and identify the performance bottlenecks. We basically face two challenges: build a performance model and integrate this model within the OCCIware framework. Our performance model is based on machine learning techniques. 

About Amadou Diarra

I’m a research and development engineer in the Efficient and Robust Distributed Systems team (ERODS) at the UGA's (University of Grenoble  Alpes) Laboratory of Informatics.  The goal of the ERODS team is to study the construction and the management of cloud-based systems. Several topics are handled in this team :
- Self-management: The goal is to integrate automation within cloud environments, with the intent to provide controllable SLA criteria for performance and robustness.
- Distributed runtime support: The goal is to provide system support to build efficient and robust cloud infrastructures.
- Virtual machine: The goal is to revisit the design of virtual machines by collapsing the virtualization layers (high-level virtual machine, middleware, operating system, and hypervisor) into one modular platform.

Before jioning the OCCIware project, I prepared a PhD on fault tolerance in distributed systems -- in particularly in P2P systems -- to solve performance issues encountered by  in ERODS team.  
So Working on OCCIware project allows me to extend this notion in cloud systems.