Research On Placement And Access Of SAAS Data

With the promotion and application of the concept of multi-tenancy in the industry, application model based on SaaS has become an efficient, advanced business application solution.Single instance multi-tenancy (Single Instance Multi-tenancy) application model reflects the features of low cost, low maintenance, and efficient application; SaaS application software providers, in order to provide better software services, have the higher demands with the SLA of application services deployment,and cloud computing emerges, just to provide a highly available system applications and highly scalable application deployment and data services.Faced to shared memory architecture and shared SaaS cloud application data, multi-tenant data is placed on the plurality of data nodes in the cloud, selecting randomly or unreasonably placed node of the cloud, the load easily lead to uneven application of the data nodes, and also consideration may increase by data access and the replica consistency;for multi-tenant cloud data access,combined with multi-tenant and cloud storage features, how to achieve multi-tenant application requests access to data and reasonable scheduling policy, then reduce the load and keep the node load balancing between nodes to achieve efficient and stable operation of the system.To solute with the problems, this paper designs the SaaS platform cloud storage architecture, based on the storage model, proposes the multi-tenant data placement and access objectives and solutions, the main contributions are as follows:1. For the problems of the cloud data for random or unreasonable place, resulting in data node load imbalance, to design a multi-tenant data placement and access algorithms for optimal multi-tenant data placement and access policies.By introducing multi-tenant data placement weighting function,build multi-tenant cloud data placement and access model.Using graph algorithms, this paper presents placement and data access algorithms and policies based on data from the nodes of the network and the data load, achieves optimal placement algorithm based on graph theory complete graph, and makes the algorithm correctness and complexity analysis to ensure that the placing node data and load balance are reasonable, reduce the costs between tenants data access and replica consistency update.2. For multi-tenant cloud data access, to design multi-tenant data access request processing functions and processes to achieve the multi-tenant cloud-oriented virtualized data access.Propose a multi-tenant cloud-oriented ata access requests model, the targets are designed to access the cloud multi-tenant data request.For different data types and access tenants types, described in detail multi-tenant cloud data access request processing, achieve the different data types of multi-tenant cloud data request processing,such as multi-tenant meta-data and business data.3. Facing the problems of multiple data nodes in the cloud requesting access and data inconsistencies,based on algorithms and models of data placement and access, dynamic scheduling access model is proposed in accordance with the state of the node load for efficient multi-tenant data access.Learn Paxos algorithm, combined with the characteristics of multi-tenant and multi-tenant metadata,introduce the metadata describing the queue data structure to solve the multi-tenant cloud data which may occur data inconsistencies and ensure the availability of SaaS applications.The paper proposed placement models and algorithms combined with multi-tenant cloud data,through experimental evaluation,which proves the correctness of multi-tenant cloud data placement and effectiveness of dynamic scheduling access.The model and policy proposed in this paper,to some extent, not only can improve the degree of shared among tenants, improve load balancing data placement, but also can reduce the segmentation copy update time, reduce the cost of the tenant access to the data for the cloud SaaS data management and provide the reference and help for us.