Nabil DJARALLAH, R&D and DevOps Director, Scalair

Nov 21 2016


How would you present the OCCIware project?

The main motivation for using Cloud platforms is their ability to provide resources according to the customer’s needs or what is referred to as elastic provisioning and deprovisioning. OCCIware is a platform that facilitates the delivery of cloud computing resources, regardless of the standards and technologies used by different Cloud platforms and through the various layers (i.e. XaaS). 

What about you and your role in the project?

At Scalair and more specifically in my department (R&D and DevOps) we work on virtual infrastructure automation (i.e. virtual machines, containers) on multi-cloud platforms and hot resources reconfiguration (Elasticity). Scalair provides IaaS resources mainly based on VMware but also on OpenStack for all project partners. Moreover, Scalair is responsible for the DCaaS (DataCenter as a Service) use-case. The use-case is based on an carpool application deployed on an elastic infrastructure (vertical and horizontal elasticity). The elasticity is managed by the OCCIware product (i.e. connector) to communicate with the hypervisors, virtual machines and containers.

What key innovation do you bring or help to develop?

Scalair brings its Cloud experience and its ability to provide resources according to the customer’s needs. Therefore, elasticity is one of the key features in Cloud computing that dynamically adjusts the amount of allocated resources to meet changes in workload demands. There are two types of elasticity: horizontal or vertical. Scalair addresses these two themes through the OCCIWARE project. Horizontal elasticity consists in adding or removing instances (Virtual Machines or VMs, Containers, etc) of computing resources. Vertical elasticity consists in increasing or decreasing characteristics of computing resources, such as cpu time, cores, memory, and Network Bandwidth. Cloud providers generally use virtualization-based approach to build their stack. VMs and Containers are the commonly virtualized resource units. Scalair brings two different Cloud platforms to the project: VMware-based and OpenStack-based testbed. ESXi (vsphere/VMware) hypervisors are the most widely used virtualization technique, they are private or property products. ESXi hypervisors allow vertical elasticity in one direction, i.e  hotplug cpu/memory add, they do not provide hotplug cpu/memory remove unless the system is rebooted or shutdown. OpenStack is a sophisticated competitor Cloud management and orchestration platform. It uses KVM/Qemu (open source hypervisors) platforms. KVM/Qemu allows hotplug add/remove of resources. Docker and Rocket are Container technologies that are widely implemented in Cloud infrastructures.  There are many powerful cluster management tools for Containers such as Kubernetes, Swarm, CoreOS. With the help of these technologies, containers can be vertically/horizontally scaled up/out or down at runtime.OpenStack can connect to container’s cluster and other virtualization management platforms such as vCenter. These technologies are very interesting for Scalair and are addressed in the project.

About Nabil DJARALLAH:
R&D and DevOps Director at the head of strategic orientations at Scalair, France. Nabil current research focuses on automation, cloud elasticity and DevOps methodologies to optimize cloud resources and customers business. He received the Ph.D degree in computer science from the University of Rennes, in France. He received also the diploma of computer science engineer from Batna University in Algeria and an M.S. degree in computer science from the University of Rennes, in France. Within the framework of his M.S. degree and Ph.D, Nabil spent almost four years at Alcatel-Lucent Bell Labs France as a research engineer. His research interests, within Bell Labs, include optimization, algorithms for constrained path computation, architectures and inter-carrier service delivery with assured QoS. After this experience, he spent more than a year at INRIA as a post-doc working on Complex Event Processing.