Data replication in large-scale data management systems

Growing popularity of large-scale applications led to generation of large volumes of data. The management of this data presents a significant challenge for service providers. In traditional systems, meeting objectives such as achieving acceptable performance and ensuring good availability are major challenges for the providers, especially when the data is distributed globally. Data replication strategies that have been proposed for traditional systems intended to maximize performance for the user. However, they are difficult to adapt to cloud systems. Meeting the performance expectations of the tenants without sacrificing the provider's profit, as well as managing resource elasticities with a pay-as-you-go pricing model, are the fundamentals challenges in this context. In this thesis, we propose a data replication strategy that satisfies the requirements of the tenant, such as performance, while guaranteeing the economic profit of the provider. Data replication is only considered if the estimated response time of database queries exceeds a threshold initially set in a contract between the provider and the tenant. Planned replication must also be economically profitable to the provider. We propose an economic model that takes into account both the expenditures and the revenues of the provider during the execution of any particular query. Once the data replication is decided to go through, a heuristic placement is used for new replicas in order to reduce the access time. In addition, a dynamic adjustment of the number of replicas is adopted to allow elastic management of resources. Proposed strategy is compared with another data replication strategy in a simulation environment. Results show that, the two compared strategies satisfied the performance objective. On the other hand, profitability of the provider is only ensured by the proposed strategy.