Study of resource management for multitenant database systems in cloud computing.

摘要

Cloud computing can offer the business community higher availability, guaranteed throughput and service rate, higher reliability, security, and cost-effective maintenance. These very attractive benefits create the growing popularity of cloud computing, and make many companies eager to migrate their existing database applications to the cloud. In the cloud, most of these applications share processors, memory, disk spaces, and databases with other companies' applications.;The cloud service providers run multiple tenants' databases in the cloud and therefore needs an effective resource sharing management method in order to allow users to keep lower overall costs without loss or degrading quality of their services. Thus, developing an efficient resource management for multi-tenant database system in cloud is essential for both business community and cloud providers. To build an efficient resource management, two key technical components that must be addressed. First, is how to create a workload aware system, this means being able to monitor resource consumption, predict a combined workload of multitenant databases and finally identify which workloads can be co-located on a database server to achieve the best consolidation and performance that is possible. The second component is having the capability to scale-out. Elastic scalability allows for the support of multiple databases and workloads, which range in size. When processing exceeds the capabilities of one node, we must be able to spread the workload across multiple nodes in order to achieve higher throughput. This thesis attempts to provide a comprehensive look at current technical issues surrounding efficient resource management of multitenant databases in cloud computing so that we are able to answer the above two questions. To achieve our goal, we will explore how a database management system operates and performs work. This will enable a better sense of the atmosphere where the multitenant databases will reside and why clouds are becoming so popular. Finally, we will present our study result of multi-tenant database, including possible architectures, current resource management issues, and proposed solutions and lastly, new ideas on how to solve these problems.