As a technique to ensure absence of undesired interference in transactional computations, Concurrency Control (CC) guarantees logical data consistency via providing transaction isolation, thus contributing to their dependability. However, single-version CC, which requires that a transaction system always works on the current version of a data item, may introduce unpredictable delays for real-time transactions because of unbounded blocking time which may cause deadline misses. Compared to single-version CC (current value of a data item is available but the historical values are overwritten and not accessible) mechanism, multi-version Concurrency Control (MVCC, historical values of a data item are maintained in a version list and accessible) mechanisms have several advantages. The benefit of multiple versions for concurrency control is helping the scheduler avoid rejecting operations, which could improve the concurrency for real-time transaction systems. Because transactions are less likely to be blocked using MVCC, timeliness could be improved. Transaction isolation levels, out of which the serializable one is the highest, control the degree of interference-freedom of concurrent transactions. Instead of serializable isolation, some MVCC mechanisms are known to achieve a relaxed level of isolation. In order to select an appropriate MVCC mechanism that guarantees both timeliness and an acceptable level of isolation for a given transaction set, trade-off analysis between isolation and timeliness is necessary. However, even though approaches have been proposed to analyze timeliness and isolation together, they only focus on lock-based single-version concurrency control algorithms, not on MVCC. In this thesis, we focus on modeling multi-version based real-time transaction system as a network of timed automata, and verify the consistency of the tradeoff transaction timeliness and isolation in UPPAAL. We propose a modular modeling approach to model real-time multi-version transaction systems by reusing and extending set of basic blocks. The proposed approach not only reduces the modeling efforts, but also enables easy adjustment for adapting current MVCC mechanism to another. Assuming a given transaction set, we model three MVCC algorithms including multi-version Timestamp Ordering, a variant of multi-version Two-Phase locking and a Two-Version Priority Ceiling Protocol, and verify both timeliness and isolation level. The verification results show that Two-Version Priority Ceiling Protocol outperforms the other two MVCC algorithms with the given transaction set.

Web services transactions have some unique characteristics. A Web transaction may be composed of a number of individual Web services, executed across multiple loosely coupled autonomous systems. Each Web service may be executed on an independent system belonging to an independent provider. There raises the question whether Web transactions can and should be maintained as a single business unit in a similar way to how transactions are maintained in classical database systems. In classical database systems, the transaction management protocol and mechanism are constrained by the primary properties of atomicity, consistency, isolation and durability (ACID). These ACID properties are the cornerstone of maintaining data integrity in transaction management. However, ACID properties were meant for centralised systems and are better suited for short transactions. Unlike short transactions, Web services transactions may be long-running; they can take hours or even days depending on the application. Composing certain actions from loosely coupled distributed business processes across multiple distributed applications is one of the essentials of Web services transactions. The classic ACID model, which is tightly coupled, is therefore seen as too rigid to support all the requirements of the new Web transactions model. The research proposes a system that increases throughput while maintaining the consistency and correctness required by the particular applications that are using the model; the system is known as AuTrA (Adaptable user-defined Transaction relaxed Approach). AuTrA allows relaxation of each ACID property. The model is adaptable to meet different situations with different characteristics. For instance, in some cases it will be appropriate to relax atomicity, whereas in others it may be appropriate to relax isolation and atomicity while maintaining consistency. The research explores how transaction support for Web services can be customised to suit the needs of varying applications and result in improved service. The AuTrA prototype has been implemented. The experimental results show that the AuTrA application is able to support the basic features of Web services transaction management, allowing users to specify their correctness requirements, and it can increase throughput of transactions in models in a flexible and reliable manner. Additional facilities allow users to specify application-specific, non-ACID criteria that can increase throughput. Safeguards have also been implemented to prevent execution of inappropriate user specifications, such as relaxation of properties that may damage data integrity. AuTrA can be used as a tool by software developers who need to compose applications from independent Web services and who wish to build applications which result in improved performance while maintaining application-required consistency.

Les logiciels adaptatifs sont une classe de logiciels qui peuvent modifier leur structure et comportement à l'exécution afin de s'adapter à des nouveaux contextes d'exécution. Le développement de logiciels adaptatifs a été un domaine de recherche très actif les dix dernières années. Plusieurs approches utilisent des techniques issues des lignes des produits afin de développer de tels logiciels. Ils proposent des outils, des frameworks, ou des langages pour construire des architectures logicielles adaptatives, mais ne guident pas les ingénieurs dans leur utilisation. De plus, ils supposent que tous les éléments spécifiés à la conception sont disponibles dans l'architecture pour l'adaptation, même s'ils ne seront jamais utilisés. Ces éléments inutiles peuvent être une cause de soucis lors du déploiement sur une cible dont l'espace mémoire est très contraint par exemple. Par ailleurs, le remplacement de composants à l'exécution reste une tâche complexe, elle doit assurer non seulement la validité de la nouvelle version, mais aussi préserver la terminaison correcte des transactions en cours. Pour faire face à ces problèmes, cette thèse propose un processus de développement de logiciels adaptatifs où les tâches, les rôles, et les artefacts associés sont explicites. En particulier, le processus vise la spécification d'informations nécessaires pour construire des architectures logicielles adaptatives. Le résultat d'un tel processus est une architecture logicielle adaptative qui contient seulement des éléments utiles pour l'adaptation. De plus, un mécanisme d'adaptation est proposé basé sur la gestion de transactions pour assurer une adaptation dynamique cohérente. Elle assure la terminaison correcte des transactions en cours. Nous proposons pour cela la notion de dépendance transactionnelle : dépendance entre des actions réalisées par des composants différents. Nous proposons la spécification de ces dépendances dans le modèle de variabilité, et de l'exploiter pour décider des fonctions de contrôle dans les composants de l'architecture, des fonctions qui assurent une adaptation cohérente à l'exécution. / Adaptive software is a class of software which is able to modify its own internal structure and hence its behavior at runtime in response to changes in its operating environment. Adaptive software development has been an emerging research area of software engineering in the last decade. Many existing approaches use techniques issued from software product lines (SPLs) to develop adaptive software architectures. They propose tools, frameworks or languages to build adaptive software architectures but do not guide developers on the process of using them. Moreover, they suppose that all elements in the SPL specified are available in the architecture for adaptation. Therefore, the adaptive software architecture may embed unnecessary elements (components that will never be used) thus limiting the possible deployment targets. On the other hand, the components replacement at runtime remains a complex task since it must ensure the validity of the new version, in addition to preserving the correct completion of ongoing activities. To cope with these issues, this thesis proposes an adaptive software development process where tasks, roles, and associate artifacts are explicit. The process aims at specifying the necessary information for building adaptive software architectures. The result of such process is an adaptive software architecture that only contains necessary elements for adaptation. On the other hand, an adaptation mechanism is proposed based on transactions management for ensuring consistent dynamic adaptation. Such adaptation must guarantee the system state and ensure the correct completion of ongoing transactions. In particular, transactional dependencies are specified at design time in the variability model. Then, based on such dependencies, components in the architecture include the necessary mechanisms to manage transactions at runtime consistently.

With the shifting landscape of computing hardware architectures and the emergence of new computing environments (e.g., large main-memory systems, hundreds of CPUs, distributed and virtualized cloud-based resources), state-of-the-art designs of transaction processing systems that rely on conventional wisdom suffer from lost performance optimization opportunities. This dissertation challenges conventional wisdom to rethink the design and implementation of transaction processing systems for modern computing environments.<div><br></div><div>We start by tackling the vertical hardware scaling challenge, and propose a deterministic approach to transaction processing on emerging multi-sockets, many-core, shared memory architecture to harness its unprecedented available parallelism. Our proposed priority-based queue-oriented transaction processing architecture eliminates the transaction contention footprint and uses speculative execution to improve the throughput of centralized deterministic transaction processing systems. We build QueCC and demonstrate up to two orders of magnitude better performance over the state-of-the-art.<br></div><div><br></div><div>We further tackle the horizontal scaling challenge and propose a distributed queue-oriented transaction processing engine that relies on queue-oriented communication to eliminate the traditional overhead of commitment protocols for multi-partition transactions. We build Q-Store, and demonstrate up to 22x improvement in system throughput over the state-of-the-art deterministic transaction processing systems.<br></div><div><br></div><div>Finally, we propose a generalized framework for designing distributed and replicated deterministic transaction processing systems. We introduce the concept of speculative replication to hide the latency overhead of replication. We prototype the speculative replication protocol in QR-Store and perform an extensive experimental evaluation using standard benchmarks. We show that QR-Store can achieve a throughput of 1.9 million replicated transactions per second in under 200 milliseconds and a replication overhead of 8%-25%compared to non-replicated configurations.<br></div>