Research On Multi-Tenant Data Index Of Saas Application Delivery Platform

With the development of Internet and the maturity of application,SaaS,a new form of software service,is gradually rising.In the SaaS model,the ISP deploy the application on its own servers,provide services for clients and charge according to the customer ordering the service number and the time.SaaS application delivery platform provides a platform for SaaS application developers, the platform shielded the characteristics of SaaS, and developers deploy applications through the platform and in the development process without having to consider SaaS features to focus on the business logic development, improve the development efficiency.SaaS applications use the "single instance multi-tenancy" approach,a number of tenants sharing a single instance,the tenants data is stored sharing in the platform databases,the multi-tenancy databases need be able to isolate the tenants data,the tenants custom its indi visual needs through the SaaS application delivery platform, the customization includs data customization,business process customization,UI customization. Therefore, the multi-tenancy database should also be able to meet the custom needs of tenants.Cloud databases pointed out the direction to the data storage of the delivery platform for SaaS application. The SaaS model is multi-tenancy shared storage architecture, resource from the database to the hardware resources can be shared in tenants,but the limited caoacity of a single data node,unable to meet the needs of growing number of tenants, therefore, the multiple data nodes is inevitable. With the globalization of their business, end-users of corporate located around the world, a single data node is unable to meet their fast response. Therefore, The SaaS model is shifting from a single-node to cloud multi data nodes.Based SaaS application delivery platform, this paper researches multi data nodes for multi-tenancy data storage.In order to meet tenants’ on-demand and efficient storage needs, multi-level indexing mechanism for multi-node storage of SaaS Application Delivery Platform is researched. The content of this study is as follows:1. Proposed Multi-level index architecture for multi-node data storage for SaaS application delivery platform. The multi-level index model use the data node index, the logic index of tenants, database physical index and efficient response to the request of the tenants. 2.Research in the third level of the index architecture, that is, tenants node index. As the data storage model for the SaaS delivery platform,multi data nodes are inevitable, meanwhile,a piece of data needs to store redundant in multiple data nodes in order to fault tolerance and load capacity of the platform. Therefore, the platform needs to route the coming tenants request to a data node which responses the request. The tenants node index could route the request to a data node to process according to the identity of the tenant in the request, and provide a transaction synchronization mechanism.3.The next study is the logic index in a single data node. Sparse tables are used to share store the data of all the tenants. Due to the particularity of the sparse table mode, creating the traditional database index in a sparse table is often impossible, so the logic index of tenants in a single data node improves the processing efficiency of the request on the basis of the physical database index.The main work of this paper is as follows:1. Propose a multi-level index architecture for the multi-tenant data storage of the SaaS application delivery platform. The multi-level index includes the tenants node index, tenants logical index, physical database index.2. Complete the establishment and maintenance mechanism of the tenants node index. The tenants node index could route the request to a data node to process according to the identity of the tenant in the request, and provide a transaction synchronization mechanism.3. Design and Implement the logical index of tenants based on the pivot table on a single data node, greatly improving the tenants’ request processing efficiency, and experimentally verify the superiority of the tenants of the logical index.4. Design and Implement of the cache optimization mechanism in the multi-level index architecture. The optimization mechanism can effectively improve the efficiency of multi-level index architecture, and verified the effectiveness of the optimization mechanism by experiments.