Font Size: a A A

The Multi-tenant Replication Data Resources Management Mechanism In Saas

Posted on:2015-03-27Degree:MasterType:Thesis
Country:ChinaCandidate:N WuFull Text:PDF
GTID:2268330431953447Subject:Computer software and theory
Abstract/Summary:PDF Full Text Request
SaaS is a software business model which employs Multi-tenant technology for the flexibility tenancy and lease-demand. As a new cloud services platform-based application delivery model, SaaS is more and more popular with its unique technical and functional advantages.In general, SaaS employs replication technology to improve redundancy in the cloud computing environment. When some copy fails, the other can continue to provide services, so as to better protect the tenant data, and improve system fault tolerance, data availability.The introduction of replication brings a series of problems.1) The traditional replication model, cannot adapt to multi-tenant architecture, or locate a copy unit of the multi-tenant data. It even cannot predict degree of multi-tenant data replication, to accommodate different tenants SLA.2) Inability to capture the characteristics of multi-tenant data, it cannot find a suitable node to place multi-tenant data to ensure good communication between the replica nodes and reduce the occurrence of a distributed transaction.3) With running multi-tenant data, the number of tenants and tenant data is increasing, the existing data copies adjustment strategy, which is difficult to adapt to the distribution of multi-tenant hotspot to share data and cannot reasonably be adjusted copy of the data to solve bottlenecks and hotspots improve data availability.To solve the above problem, this paper discusses the multi-tenant replication model in SaaS software delivery platform, and proposes a dynamic replication management model for multi-tenant, the degree of replica projection mechanism, and placing and adjustment mechanism. The main work is as follows:(1) For multi-tenant data storage characteristics, it proposes a management model oriented multi-tenant, to solve replication management problems based on big wide table.It puts forward a dynamic replication management model for multi-tenant, and designs a replica management model framework based on SaaS Software Delivery Platform. It describes the relationship between the various components of the model as well as a brief description of the multi-tenant data access. It also improves traditional distributed network topologies, and proposes a hierarchical network model to solve the bottleneck of a single point and hot problem.(2) For request message distribution of multi-tenant data, it puts forward replication prediction based on message queue to improve the accuracy and efficiency of multi-tenant replication management.It analyzes the data node failure probability in multi-tenant cloud platform environment, and calculates the minimum replication degree of the initial threshold platforms to ensure data availability. It researches the distribution of multi-tenant data request message, derived message probability distribution. It puts forward the message queue model oriented by multi-tenant data request message distribution, and calculates the optimal replication number.(3) On the foundation of dynamic multi-tenant-oriented management model, it puts forward multi-tenant data replica placement and adjustment algorithm, to reduce blocking communication between nodes and the distributed transaction.It extracts the factors in replication data placement and adjustment based on message queue to calculate the appropriate node position. At last, through message filtering, message queues, and message listener, it proposes a three-stage adjustment to ensure timely adjustments to reduce the probability of effect data availability.The paper puts forward a dynamic replication management model. It can forecast the number of copies, coordinate placement replication data, and adjust copy of data on time. It can reduce the probability of effect tenant availability, tenants distributed transaction, and inter-node communication costs, to conserve system resources, balancing the system load, and improve tenant data availability.
Keywords/Search Tags:SaaS, Multi-tenant, Replication Prediction, Replication Placement, Replication Adjustment
PDF Full Text Request
Related items