We are witnessing exponential growth in the number of powerful, multiply-connected, energy-rich stationary and mobile nodes, which will make available a massive pool of computing and communication resources. We claim that cloud computing can provide resilient on-demand computing, and more effective and efficient utilization of potentially infinite array of resources. Current cloud computing systems are primarily built using stationary resources. Recently, principles of cloud computing have been extended to the mobile computing domain aiming to form local clouds using mobile devices sharing their computing resources to run cloud-based services. However, current cloud computing systems by and large fail to provide true on-demand computing due to their lack of the following capabilities: 1) providing resilience and autonomous adaptation to the real-time variation of the underlying dynamic and scattered resources as they join or leave the formed cloud; 2) decoupling cloud management from resource management, and hiding the heterogeneous resource capabilities of participant nodes; and 3) ensuring reputable resource providers and preserving the privacy and security constraints of these providers while allowing multiple users to share their resources. Consequently, systems and consumers are hindered from effectively and efficiently utilizing the virtually infinite pool of computing resources. We propose a platform for mobile cloud computing that integrates: 1) a dynamic real-time resource scheduling, tracking, and forecasting mechanism; 2) an autonomous resource management system; and 3) a cloud management capability for cloud services that hides the heterogeneity, dynamicity, and geographical diversity concerns from the cloud operation. We hypothesize that this would enable 'Collaborative Computing Cloud (C3)' for on-demand computing, which is a dynamically formed cloud of stationary and/or mobile resources to provide ubiquitous computing on-demand. The C3 would support a new resource-infinite computing paradigm to expand problem solving beyond the confines of walled-in resources and services by utilizing the massive pool of computing resources, in both stationary and mobile nodes. In this dissertation, we present a C3 management platform, named PlanetCloud, for enabling both a new resource-infinite computing paradigm using cloud computing over stationary and mobile nodes, and a true ubiquitous on-demand cloud computing. This has the potential to liberate cloud users from being concerned about resource constraints and provides access to cloud anytime and anywhere. PlanetCloud synergistically manages 1) resources to include resource harvesting, forecasting and selection, and 2) cloud services concerned with resilient cloud services to include resource provider collaboration, application execution isolation from resource layer concerns, seamless load migration, fault-tolerance, the task deployment, migration, revocation, etc. Specifically, our main contributions in the context of PlanetCloud are as follows. 1. PlanetCloud Resource Management • Global Resource Positioning System (GRPS): • Global mobile and stationary resource discovery and monitoring. A novel distributed spatiotemporal resource calendaring mechanism with real-time synchronization is proposed to mitigate the effect of failures occurring due to unstable connectivity and availability in the dynamic mobile environment, as well as the poor utilization of resources. This mechanism provides a dynamic real-time scheduling and tracking of idle mobile and stationary resources. This would enhance resource discovery and status tracking to provide access to the right-sized cloud resources anytime and anywhere. • Collaborative Autonomic Resource Management System (CARMS): Efficient use of idle mobile resources. Our platform allows sharing of resources, among stationary and mobile devices, which enables cloud computing systems to offer much higher utilization, resulting in higher efficiency. CARMS provides system-managed cloud services such as configuration, adaptation and resilience through collaborative autonomic management of dynamic cloud resources and membership. This helps in eliminating the limited self and situation awareness and collaboration of the idle mobile resources. 2. PlanetCloud Cloud Management Architecture for resilient cloud operation on dynamic mobile resources to provide stable cloud in a continuously changing operational environment. This is achieved by using trustworthy fine-grained virtualization and task management layer, which isolates the running application from the underlying physical resource enabling seamless execution over heterogeneous stationary and mobile resources. This prevents the service disruption due to variable resource availability. The virtualization and task management layer comprises a set of distributed powerful nodes that collaborate autonomously with resource providers to manage the virtualized application partitions. / Ph. D.

Universiẗat, Diss.--Bayreuth, 2006.

Twenty-first century Field-Programmable Gate Arrays (FPGAs) are no longer used for implementing simple “glue logic” functions. They have become complex arrays of reconfigurable logic resources and memories as well as highly optimised functional blocks, capable of implementing large systems on a single chip. Moreover, Dynamic Partial Reconfiguration (DPR) capability permits to adjust some logic resources on the chip at runtime, whilst the rest are still performing active computations. During the last few years, DPR has become a hot research topic with the objective of building more reliable, efficient and powerful electronic systems. For instance, DPR can be used to mitigate spontaneously occurring bit upsets provoked by radiation, or to jiggle around the FPGA resources which progressively get damaged as the silicon ages. Moreover, DPR is the enabling technology for a new computing paradigm which combines computation in time and space. In Reconfigurable Computing (RC), a battery of computation-specific circuits (“hardware tasks”) are swapped in and out of the FPGA on demand to hold a continuous stream of input operands, computation and output results. Multitasking, adaptation and specialisation are key properties in RC, as multiple swappable tasks can run concurrently at different positions on chip, each with custom data-paths for efficient execution of specific computations. As a result, considerable computational throughput can be achieved even at low clock frequencies. However, DPR penetration in the commercial market is still testimonial, mainly due to the lack of suitable high-level design tools to exploit this technology. Indeed, currently, special skills are required to successfully develop a dynamically reconfigurable application. In light of the above, this thesis aims at bridging the gap between high-level application and low-level DPR technology. Its main objective is to develop Operating System (OS)-like support for high-level software-centric application developers in order to exploit the benefits brought about by DPR technology, without having to deal with the complex low-level hardware details. The developed solution in this thesis is named as R3TOS, which stands for Reliable Reconfigurable Real-Time Operating System. R3TOS defines a flexible infrastructure for reliably executing reconfigurable hardware-based applications under real-time constraints. In R3TOS, the hardware tasks are scheduled in order to meet their computation deadlines and allocated to non-damaged resources, keeping the system fault-free at all times. In addition, R3TOS envisages a computing framework whereby both hardware and software tasks coexist in a seamless manner, allowing the user to access the advanced computation capabilities of modern reconfigurable hardware from a software “look and feel” environment. This thesis covers all of the design and implementation aspects of R3TOS. The thesis proposes a novel EDF-based scheduling algorithm, two novel task allocation heuristics (EAC and EVC) and a novel task allocation strategy (called Snake), addressing many RC-related particularities as well as technological constraints imposed by current FPGA technology. Empirical results show that these approaches improve on the state of the art. Besides, the thesis describes a novel way to harness the internal reconfiguration mechanism of modern FPGAs to performinter-task communications and synchronisation regardless of the physical location of tasks on-chip. This paves the way for implementing more sophisticated RC solutions which were only possible in theory in the past. The thesis illustrates R3TOS through a proof-of-concept prototype with two demonstrator applications: (1) dependability oriented control of the power chain of a railway traction vehicle, and (2) datastreaming oriented Software Defined Radio (SDR).

Cette thèse étudie la gestion de ressources pour des services de calcul intensif à la demande sur une grappe de calcul partagée. Dans un tel contexte, il s'agissait de définir des outils d'exploitation qui permettent d'allouer dynamiquement les ressources pour l'exécution des requêtes à la demande, de partager équitablement les ressources entre les différents services, tout en maximisant leur utilisation. Financé par le pôle de compétitivité Minalogic dans le cadre du projet Ciloe (http://ciloe.minalogic.net), ce travail s'adresse à des organisations de types PME ou PMI, où les budgets de fonctionnement ne permettent pas de supporter les charges d'une infrastructure de calcul dédiée. Dans un premier temps, nous avons dressé un état de l'art sur la gestion de ressources dans les domaines de nuage de calcul et de calcul intensif. Puis, tirant partie de cette étude, nous avons défini une architecture virtualisée pour faciliter l'exécution dynamique des requêtes grâce à un gestionnaire de ressources spécifique. Nous avons enfin proposé une stratégie de partage et d'allocation de ressources flexible qui offre un compromis entre équité et utilisation efficace de ressources. Ayant travaillé dans un contexte de collaboration avec des industriels, nous avons développé un prototype comme une preuve de concept. Basé sur des standards ouverts, ce prototype s'appuie sur des outils existants de virtualisation tel que OpenNebula pour allouer et manipuler les machines virtuelles sur les noeuds de la grappe. A partir de ce prototype et diverses charges de travail qui sont détaillés dans cette thèse, nous avons mené des expériences pour évaluer l'architecture et les algorithmes de gestion de ressources. Les résultats montrent que ces différentes contributions satisfont les objectifs fixés tout en étant performantes et efficaces. / This thesis studies resource management for on-demand computing services through a shared cluster. In such a context, the aim was to propose tools to enable allocating resources automatically for executing on-demand user requests, to enable sharing resources proportionally among those services, while maximizing their use. Funded by the Minalogic global business cluster through the Ciloe Project (http://ciloe.minalogic.net), this work targets on organizations such as SMB, which are not able to support the charge of purchasing and maintaining a dedicated computing infrastructure. Firstly, we have achieved a deep survey in the areas of on-demand computing and high performance computing. From this survey, we have defined a virtualized architecture to enable dynamic execution of user requests thanks to a special resource manager. Finally, we have proposed policies and algorithms which are so flexible to offer a suitable tradeoff between equity and resource use. Having worked in a context of industrial collaboration, we have developed a prototype of our proposal as a proof of concept. Based on open standards, this prototype relies on existing virtualization tools such as OpenNebula for allocating and manipulating virtual machines over the cluster's nodes. From this prototype along with various workloads, we have carried out experiments to evaluate our architecture and scheduling algorithms. Results have shown that our contributions allow to achieve the expected goals while being reliable and efficient.