Session Open Source Cloud

Session Chair


A plate-form for monitoring of Cloud infrastructure and applications

Speaker: Bernard Boltz, Orange
Abstract: Various factors introduce specific and new constrains for the monitoring of the Cloud infrastructure and services. For instance, a monitoring tool has to adapt to the quick and elastic provisioning, an important feature of the Cloud. We developed from an open source platform, Icinga / Nagvis, several modules that allow: 1) For the data center infrastructure monitoring - Auto discovery of components of virtualized infrastructures and automatic set-up of their monitoring according to appropriate templates - Dialogue with the VMware API to query, among other things, the status of clusters, resource pools - Maps management in order to drill up or down in the hierarchy of the data center et the virtualized components. 2) For the application monitoring - Management of an applications dashboard that contains a representative set of key performance indicators (KPIs). - Calculation of the indicators - Management of maps showing the links between application and infrastructure. This implementation allows enhanced monitoring. We have also learned the lessons of the experimentation. The presentation will present these modules and limitations encountered. 

OpenCloudware: the Cloud application lifecycle management platform

Speaker: Alexandre Lefebvre, Orange, OW2 ; Thomas Debru, UShareSoft
Abstract: The OpenCloudware collaborative R&D project aims at building an open software engineering platform, for the collaborative development of distributed applications to be deployed on multiple Cloud infrastructures. It will be available through a self-service portal. We target virtualized multi-tier applications such as JavaEE - OSGi. The results of OpenCloudware contain a set of software components to manage the lifecycle of such applications, from modelling (Think), developing and building images (Build), to a multi-IaaS compliant PaaS platform (Run) for their deployment, orchestration, performance testing, self-management (elasticity, green IT optimisation) and provisioning. Applications can be deployed potentially on multiple IaaS (supporting either one IaaS at a time, or hybrid scenarios). The results of the project are gradually made available as open source components through the OW2 Open Source Cloudware initiative. The OpenCloudware project is coordinated by France Telecom Orange, funded by the french Fonds National pour la Société Numérique (FSN) and endorsed by competitiveness clusters Minalogic, Systematic and SCS. The project runs between January 2012 and December 2014.

Service4All: An Open Source Solution for Service Oriented Applications over Clouds

Speaker: Sun Hailong, Beihang University
Abstract: In this presentation, I will introduce Service4All, our open source work focusing on the PaaS layer of cloud computing. There are three building blocks with Service4All including ServiceXchange, Service Foundry and Service Oriented App Engine. ServiceXchange is an online service repository, which is responsible for providing reusable services for building service-oriented applications; Service Foundry is an online application development environment, which supports building composite services and mashups; Service Oriented App Engine provides a dependable and transparent hosting platform for running applications built with Service Foundry. In all, the three building blocks work together to provide fully-fledged support for all about service-oriented application development. I will also give a short demo of how Service4All works.

CompatibleOne the Open Source Cloud Broker

Speaker:Jamie Marshall, Prologue
Abstract:Next generation cloud service management software will offer service brokering capabilities. This session will provide an overview of the architecture of an open source cloud service broker that would allow for the description, provision, deployment and management of any type or configuration of cloud services delivered by heterogeneous cloud service providers in accordance with a consumer's SLA. A broker allows the federation of resources across heterogeneous Cloud Service Providers and thus plays an important role in avoiding vendor lock-in. The session will end with a live demonstration showing how a federation of multiple OpenStack providers can be managed by CompatibleOne, an open source cloud service broker, in a multi-cloud environment.

OpenPaaS: the open-source Enterprise Social network for the Cloud

Speaker: Jean-Pierre Lorré, Linagora
Abstract: The OpenPaaS project aims at developing an Open-Source Enterprise Social Network that leverage OBM collaborative solution and Petals BPM workflow engine and run-time. Leaded by LINAGORA, the French Open-Source Software leader, OpenPaaS addresses the whole collaborative application life-cycle. It provides basic features such as profile definition, collaborative mail, agenda, forms and workflow editors as well as an open API for external contributors. OpenPaaS’s run-time reuses Petals BPEL process engine and Petals Enterprise Service Bus. These tools are deployed on an hybrid IaaS infrastructure that allows to connect heterogeneous external applications. OpenPaaS project is funded by the FSN (Fond National pour la Société Numérique). 

VENUS-C: lessons learned of using open source clouds in eScience

Speaker: Andrea Manieri, Engineering Group
Abstract: VENUS-C is the first European distributed computing infrastructure to adopt a User-centric approach to Cloud Computing, putting the needs of end-User communities of researchers and small businesses at the forefront of development. A representative User community sample was composed by the 7 scenarios, the 15 pilots and the 5 experiments from eleven different disciplines and SMEs. Chief among the success stories is Green Prefab, a start-up company for architects and civil engineers, which hosts a new innovation cluster stemming from VENUS-C, HUB-Engineering ( The VENUS-C Architecture Report describes the blueprint for the platform based on which a reference implementation has been done applying a rapid prototyping approach with interactive releases and User validation. The project publicly released open-source components (, namely: • Generic Worker (PMES (Process Management Enactment Services) and programming framework) • COMPSs (PMES and programming framework) • CDMI Proxy • Accounting and billing service • Traffic Redundancy Elimination (TRE) • Interoperability Toolkit As a result, it is clear that the adoption of Cloud and Virtualisation technologies and services is a win-win situation for the scientific community, as it releases benefits for both research institutions, industry. 

XLcloud: Remote rendering for games and 3D graphics applications in the cloud

Speakers: Patrick Petit, Bull & Marius Preda, Telecom SudParis
Abstract: The XLcloud project proposes a cutting-edge high performance cloud computing environment that enables low-latency remote rendering of 3D and complex 2D applications such as online gaming, cultural artifacts visualization and virtual body inspection in the cloud.
Many people refer to cloud computing as managing virtual machines that run on top of an hypervisor. However, in certain circumstances, using virtualization is not desirable when applications have stringent performance requirements or need access to devices that are not supported by hypervisors like GPUs for example.
However, it is still very convenient to utilize the features of a IaaS cloud such as instance management and image management for use cases that require the automated provisioning of a collection of cloud resources. As such, the main paradigm of XLcloud is based upon the ability to instantiate purpose-built virtual cluster instances that feature a collection of connected physical and virtual resources to support high-performance / low-latency video encoding and interactivity streaming artifacts.
The presentation of XLcloud will provide a description of the Remote Rendering Virtual Cluster architecture with a particular emphasis on server components that are designed to enable a highly pervasive environment allowing end-users to run their 3D applications anywhere on any device. In addition, we will describe a low computational method to perform ROI-based (Region Of Interest) video encoding and adaptive streaming as well as a multi-modal representation of the images and videos (i.e. pixel vs graphics) to reduce the bit rate requirements. Finally, we will conclude with a short description of a partitioning algorithm and some implementation details.

BtrPlace: An Extensible VM manager to face up to SLA expectations in a cloud

Speaker: Fabien Hermenier, University of Nice Sophia Antipolis and Jean-Marc Menaud, Ecoles des Mines de Nantes
Abstract: Clients of IaaS providers are looking for dependable infrastructures that can cope with their SLA requirements. To stay attractive, a cloud must then rely on a VM placement algorithm with features matching as possible the SLAs expectations. These constraints are however very specific to each of the tenants but also the infrastructure. They also cover a large range of concerns (reliability, performance, security, energy-efficiency, ...) that are continuously evolving according to new trends and new technologies. Current VM placement algorithms are designed as monolithic systems. Clouds administrators and clients are then passive actors, waiting for the future release of the placement algorithm that may match their current expectations. During this presentation, we will present BtrPlace, an open source VM manager that can be specialized on demand by third party developers to integrate new placement constraints. BtrPlace has already been used successfully to address high-availability, security but also energy-efficiency concerns. It has been developped since 2006 in the Mines de Nantes then with the University of Nice-Sophia Antipolis. It is now a part of the OpenCloudware project that aims at building an open platform for the development and the deploymet of distributed application on clouds. 


Speaker: Alban Richard, UShareSoft


Playing with SOA, EDA and Event processing in the Cloud

Speaker:Christophe Hamerling, Linagora
Abstract: The PLAY project ( is developing an elastic and reliable architecture for dynamic and complex, event-driven interaction in large highly distributed and heterogeneous service systems. Such an architecture will enable ubiquitous exchange of information between heterogeneous services, providing the possibilities to adapt and personalize their execution, resulting in the so-called situational-driven process adaptivity.

ETICS: quality certification of open source cloud software

Speaker: Gabriele Giammatteo, Engineering Group
Abstract: Cloud Computing is becoming the new paradigm for the Internet and Service provisioning. Open source solutions, especially those resulting from R/D projects, often are neglected due to the absence of track-records in use by real world applications. ETICS is a build, test and certification software developed since 2004 and up to now used at CERN, ENG and others to build and test various complex software systems, such us glite ( and gcube ( ETICS will be used in the Ocean support action to certify the quality of RTD open source cloud software. Ocean aims at identifying and tracking open-source cloud development projects in an open online portal to promote cross-project collaboration. Ocean will also provide projects with an on-demand build and test service reducing their need for installing and operating tools, and a beta-tester service giving them back a comprehensive review and feedback. This presentation provides an overview of the ETICS approach and tool and the way it's planned to be adopted within the Ocean project.

The Sirocco multi-cloud management framework

Speaker: Frederic Dang Tran, Orange Labs
Abtsract: While private cloud solutions as well as public cloud service offers are numerous, effective management still remains a challenge where it comes to multi-cloud or hybrid cloud deployment models where applications span several clouds. This presentation gives an overview of the OW2 Sirocco multi-cloud management framework with a particular focus on two topics: API mediation and placement management. Sirocco allows consumers to provision resources on multiple cloud providers whether private or public and relies on an orchestration layer that hides the heterogeneity of the underlying providers (in terms of API, virtualization technology…) through a unified user interface and API. Sirocco has adopted the DMTF Cloud Infrastructure Management Interface (CIMI) whose 1.0 release was released on 29 August 2012. Internally, this unified API is translated to the specific API and virtual resource model of each provider. Sirocco currently supports Amazon EC2, VMware vCloud Director as well as OpenStack as back-end clouds. A key component of Sirocco focuses on provider selection and placement management, namely how to select a cloud provider that offers the best match between the client request (QoS, geographical constraints…) and the provider’s capabilities. Sirocco integrates a flexible decision-making framework that relies on a rule-based engine. Second speaker: Pawel Rubach Orange Labs Poland