The Multi-granular Locking Mechanism Research Of XML Database Based On Pattern Tree

With the continuous growth of the network bandwidth, the cloud computing is gettingmore and more attention and application. DaaS (Database as a Service) is a data-managementmodel of the third-party hosting services, and a new database application in the cloud storagesystem. The DAAS provides customers the services to create, store, and access the database onthe host site. The documents expressed in XML format have fast development and wideapplication because of the scalable and self-descriptive characteristic. With the development ofDaaS, the XML data processing techniques have been behind the new situation in some respects.High efficiency and low cost of processing XML data in the DaaS environment has become abig problem. The locking mechanism has a very important role in XML database transactionprocessing, and the existing mechanism of the blockade needs to be improved because theycannot meet the requirements in DaaS.This article intends to design a new mechanism of concurrency control and multi-queryoptimization to improve the performance of XML database. On technology strategy, it uses adynamic model tree based on XML query algebra to build a multi-granularity lockingmechanism, and reduce the overhead of predicate processing by interval histogram method. Itdigs the subset of the "critical path" in more than one pattern tree, and optimizes theperformance of multi-XML query by query rewriting. By constructing the query view based ondynamic global schema tree and XML transaction processing coordinator, it deals with theproblem of multi-query optimization and concurrency control in the DAAS mode. On this basis,the article designs a prototype system oriented the DAAS application and builds the appropriateevaluation system.