Vers une méthodologie d'optimisation du placement des objets DBAAS dans un environement de Cloud computing

Résumé

DBaaS cloud providers offer the database as service. The growth and highly competitive nature of this economy has compelled them to optimize the use of their data centers, in order to offer attractive services at a lower cost. From this perspective, Several technologies and techniques have been designed to optimize costs such as the storage cost (e.g. the use of hybrid storage systems), the computing cost (e.g. virtualization) and the maintenance cost (e.g. multi-tenant). In this thesis, we are interested in hybrid storage system HDD-SSD. Indeed, Hard Disk Drives (HDD) represent energy-intensive an inefficient devices compared to compute units. However, their low cost per gigabyte and their long lifetime may constitute positive arguments. Unlike HDD, flash-based Solid-State Disks (SSD) are more efficient and consume less power, but their high cost per gigabyte and their short lifetime (compared to HDD) represent major constraints. The idea is to combine SSD and HDD seeking the performance of SSDs with the price of HDDs. However, adopting such a technology without the use of a object placement strategy suitable for hybrid storage systems will only increase the costs without necessarily leading to improved performance. It is therefore necessary to design placement strategies that manage the placement of objects and decide where and when to place an object in a storage class. These strategies must be able to manage the movement of data between different storage classes in the case of workloads fluctuating. In this work, we take relational databases as a case study. However, our approaches are applicable for NoSQL databases. In this thesis we propose a new autonomic model for object placement optimization based on MAPE-K (Monitor, Analyze, Plan, Execute, Knowledge) whereby in addition to the key aspects of Cloud computing paradigm, the Object I/O and related storage systems are considered. Our first contribution consist to propose a cost model which takes into account the object I/O profile, the storage system characteristics, and the cloud environment constraints. The second contribution consist to propose a Cost based Object Placement Strategies (COPS) on HSS for relational cloud DBMSs. We propose two strategies. The first one is a meta-heuristic approach based on a genetic algorithm (G-COPS). As metaheuristic algorithms are not always the best answer for optimization problems, we also propose a specialized heuristic-based solution (H-COPS). The idea of H-COPS consists in computing a first object placement solution and enhancing it incrementally based on the characteristics of the I/O workloads and customer penalties.

Citer ce document

Accès au document

Ce document est hébergé sur son dépôt institutionnel d'origine.

Auteur(s)

Statistiques

Consultations : 3

Téléchargements : 0