高级检索

    基于Pareto最优的DaaS数据布局策略

    A Pareto-Based Data Placement Strategy in Database as a Service Model

    • 摘要: 数据库即服务(database as a service, DaaS)作为一种新型的数据存储提供模式被广泛应用.随着大数据时代的到来,数据量急剧增加,DaaS模式下的数据布局问题显得更加重要,即服务提供商如何根据应用中不同数据的性能需求对数据进行合理布局,将会对提高服务质量、增强用户体验和降低自身服务成本产生重要影响.然而对于服务提供者来说提高服务质量和降低服务成本是一对矛盾的目标.提出DaaS模式下的数据布局图概念,应用Pareto最优思想适合于解决多目标矛盾性问题的特点,给出一个基于性能-代价均衡的多节点DaaS数据布局策略.通过与随机策略和贪婪策略等传统策略的实验比较,方法能保证DaaS服务提供商用尽可能少的代价为用户提供更好的服务质量,实现服务质量与资源代价两个目标的均衡.

       

      Abstract: DaaS (database as a service) as a new data provisioning pattern, has been widely applied. With the arrival of the era of big data, the amount of data is increasing dramatically and the data placement problem in DaaS model becomes more and more important. How to place the data according to their different performance requirements has an important effect on improving the quality of service, enhancing user experience and reducing the service cost. However, for service providers, improving service quality and reducing service cost are contradictory. It can't use less cost to buy the higher performance servers so that it is difficult to provide better service quality for users. It models the DaaS data placement problem as the DaaS data placement graph. Then using the pareto optimal thought, it puts forward a DaaS data placement strategy based on performance-cost equilibrium, including the initial solutions generation stage, multi-objective optimization stage and the final solution generation stage. It supports that DaaS service providers use less cost to provide the better service quality for users. Compared with the random strategy and greedy strategy through experiments, the strategy proposed in this paper can effectively reduce the resources cost while provide better service quality. It achieves the trade-off of the service performance and resources cost.

       

    /

    返回文章
    返回