With the widespread use of the Internet, the popularity of web applications has significantly increased. Such applications are subject to unpredictable workload conditions that vary from time to time. For example, an e-commerce website may face higher workloads than normal during festivals or promotional schemes. Such applications are critical and performance related issues, or service disruption can result in financial losses. Cloud computing with its attractive feature of dynamic resource provisioning (elasticity) is a perfect match to host such applications. The rapid growth in the usage of cloud computing model, as well as the rise in complexity of the web applications poses new challenges regarding the effective monitoring and management of the underlying cloud computational resources. This thesis investigates the state-of-the-art elastic methods including the models and techniques for the dynamic management and provisioning of cloud resources from a service provider perspective. An elastic controller is responsible to determine the optimal number of cloud resources, required at a particular time to achieve the desired performance demands. Researchers and practitioners have proposed many elastic controllers using versatile techniques ranging from simple if-then-else based rules to sophisticated optimisation, control theory and machine learning based methods. However, despite an extensive range of existing elasticity research, the aim of implementing an efficient scaling technique that satisfies the actual demands is still a challenge to achieve. There exist many issues that have not received much attention from a holistic point of view. Some of these issues include: 1) the lack of adaptability and static scaling behaviour whilst considering completely fixed approaches; 2) the burden of additional computational overhead, the inability to cope with the sudden changes in the workload behaviour and the preference of adaptability over reliability at runtime whilst considering the fully dynamic approaches; and 3) the lack of considering uncertainty aspects while designing auto-scaling solutions. This thesis seeks solutions to address these issues altogether using an integrated approach. Moreover, this thesis aims at the provision of qualitative elasticity rules. This thesis proposes a novel biologically-inspired switched feedback control methodology to address the horizontal elasticity problem. The switched methodology utilises multiple controllers simultaneously, whereas the selection of a suitable controller is realised using an intelligent switching mechanism. Each controller itself depicts a different elasticity policy that can be designed using the principles of fixed gain feedback controller approach. The switching mechanism is implemented using a fuzzy system that determines a suitable controller/- policy at runtime based on the current behaviour of the system. Furthermore, to improve the possibility of bumpless transitions and to avoid the oscillatory behaviour, which is a problem commonly associated with switching based control methodologies, this thesis proposes an alternative soft switching approach. This soft switching approach incorporates a biologically-inspired Basal Ganglia based computational model of action selection. In addition, this thesis formulates the problem of designing the membership functions of the switching mechanism as a multi-objective optimisation problem. The key purpose behind this formulation is to obtain the near optimal (or to fine tune) parameter settings for the membership functions of the fuzzy control system in the absence of domain experts’ knowledge. This problem is addressed by using two different techniques including the commonly used Genetic Algorithm and an alternative less known economic approach called the Taguchi method. Lastly, we identify seven different kinds of real workload patterns, each of which reflects a different set of applications. Six real and one synthetic HTTP traces, one for each pattern, are further identified and utilised to evaluate the performance of the proposed methods against the state-of-the-art approaches.

Le Cloud Computing ou "informatique en nuage" est un nouveau paradigme émergeant pour l’exploitation des services informatiques distribuées à large échelle s’exécutant à des emplacements géographiques répartis. Ce paradigme est de plus en plus utilisé pour le déploiement et l’exécution des applications en général et des applications à base de services en particulier. Les applications à base de services sont décrites à l’aide du standard Service Component Architecture (SOA) et consistent à inter-lier un ensemble de services élémentaires et hétérogènes en utilisant des spécifications de composition de services appropriées telles que Service Component Architecture (SCA) ou encore Business Process Execution Language (BPEL). Provisionner une application dans le Cloud consiste à : (1) allouer les ressources dont elle a besoin pour s’exécuter, (2) déployer ses sources sur les ressources allouées et (3) démarrer l’application. Cependant, les solutions Cloud existantes sont limitées en termes de plateformes d’exécution. Ils ne peuvent pas toujours satisfaire la forte hétérogénéité des composants des applications à base de services. Pour remédier à ces problèmes, les mécanismes de provisioning des applications dans le Cloud doivent être reconsidérés. Ces mécanismes doivent être assez flexibles pour supporter la forte hétérogénéité des composants sans imposer de modifications et/ou d’adaptations du côté du fournisseur Cloud. Elles doivent également permettre le déploiement automatique des composants dans le Cloud. Si l’application à déployer est mono-composant, le déploiement est fait automatiquement et de la même manière, et ce quelque soit le fournisseur Cloud choisi. Si l’application est à base de services hétérogènes, des fonctionnalités appropriées doivent être mises à la disposition des développeurs pour qu’ils puissent définir et créer les ressources nécessaires aux composants avant de déployer l’application. Dans ce travail, nous proposons une approche appelée SPD permettant le provisioning des applications à base de services dans le Cloud. L’approche SPD est constituée de 3 étapes : (1) découper des applications à base de services en un ensemble de services élémentaires et autonomes, (2) encapsuler les services dans des micro-conteneurs spécifiques et (3) déployer les micro-conteneurs dans le Cloud. Pour le découpage, nous avons élaboré un ensemble d’algorithmes formels assurant la préservation de la sémantique des applications une fois découpées. Pour l’encapsulation, nous avons réalisé des prototypes de conteneurs de services permettant l’hébergement et l’exécution des services avec seulement le minimum des fonctionnalités nécessaires. Pour le déploiement, deux cas sont traités i.e. déploiement sur une infrastructure Cloud (IaaS) et déploiement sur une plateforme Cloud (PaaS). Pour automatiser le processus de déploiement, nous avons défini : (i) un modèle de description des ressources unifié basé sur le standard Open Cloud Computing Interface (OCCI) permettant de décrire l’application et ses ressources d’une manière générique quelque soit la plateforme de déploiement cible et (ii) une API appelée COAPS implémentant ce modèle et permettant de l’approvisionnement et la gestion des applications en utilisant des opérations génériques quelque soit la plateforme cible / Cloud Computing is a new supplement, consumption, and delivery model for IT services based on Internet protocols. It is increasingly used for hosting and executing applications in general and service-based applications in particular. Service-based applications are described according to Service Oriented Architecture (SOA) and consist of assembling a set of elementary and heterogeneous services using appropriate service composition specifications like Service Component Architecture (SCA) or Business Process Execution Language (BPEL). Provision an application in the Cloud consists of allocates its required resources from a Cloud provider, upload source codes over their resources before starting the application. However, existing Cloud solutions are limited to static programming frameworks and runtimes. They cannot always meet with the application requirements especially when their components are heterogeneous as service-based applications. To address these issues, application provisioning mechanisms in the Cloud must be reconsidered. The deployment mechanisms must be flexible enough to support the strong application components heterogeneity and requires no modification and/or adaptation on the Cloud provider side. They also should support automatic provisioning procedures. If the application to deploy is mono-block (e.g. one-tier applications), the provisioning is performed automatically and in a unified way whatever is the target Cloud provider through generic operations. If the application is service-based, appropriate features must be provided to developers in order to create themselves dynamically the required resources before the deployment in the target provider using generic operations. In this work, we propose an approach (called SPD) to provision service-based applications in the Cloud. The SPD approach consists of 3 steps: (1) Slicing the service-based application into a set of elementary and autonomous services, (2) Packaging the services in micro-containers and (3) Deploying the micro-containers in the Cloud. Slicing the applications is carried out by formal algorithms that we have defined. For the slicing, proofs of preservation of application semantics are established. For the packaging, we performed prototype of service containers which provide the minimal functionalities to manage hosted services life cycle. For the deployment, both cases are treated i.e. deployment in Cloud infrastructure (IaaS) and deployment in Cloud platforms (PaaS). To automate the deployment, we defined: (i) a unified description model based on the Open Cloud Computing Interface (OCCI) standard that allows the representation of applications and its required resources independently of the targeted PaaS and (ii) a generic PaaS application provisioning and management API (called COAPS API) that implements this model