| Information is one of the most substantial resourses obtained by computer. While the modern information technology and Internet technology are booming, people have the great wish to share information resources to the greatest extend and to obtain and utilize them in a more efficient way. Therefore,"Information Integration" has been one of the most important topic that get lots of attention from the academia in the field of information science. However, present information integration methods and technologies that stem from solving specific and single integration needs, fail to accommodate to the complex integration requirement and the expanding integration scope in network environment. So, we have to consider a new way to deal with information integration to adapt to the scalization, initiative and intensification trends of information integration.According to above cognition of information integration, we propose to research a new information integration method in SaaS (Software as a Service) mode and corresponding technologies in this thesis. Three key problems are focused in our research work:(1) How to provide low-cost and transparent storage and management infrastructure for information data, especially the relational data?(2) How to organize and manage large amounts of information services provided by different users and ensure the availability of these information services?(3) How to operate and optimize the plan of information resources, the governance of information services and the schedule of the integration applications?Aiming at above problems, we summarize the stage of art and trend of information integration first in this thesis. And then, based on the analyses of open mode information integration, we propose a SaaS mode information integration method, called SMII, and a set of technology including multi-tenant database services, information service community, flexible monitor for service availability and three-level metadata management. Finally, it is given that the design, implementation and application of a SaaS mode information integration platform.The thesis mainly contributed to the work that:(1) To solve the problems of traditional closed, passive and inefficient methods for information integration, a SaaS mode information integration method (named as SMII) and corresponding information integration reference framework are proposed to adapt to the scalization, initiative and intensification trends of information integration. In SMII method, software services are provided at data level, service level and application level respectively, to support the three core procedures of information resource sharing, information service provision and information application development. According to the real practice and compared to the current typical methods for information integration, SMII method shows the advantage of openness in information integration scope, high extent in end-user participation of information integration and cost-effectiveness in integration application development, and is very suitable for the situation of information integration scopes extending and requirements changing in Web2.0age.(2) Focused on the problem of multi-tenant database service, we put forward a virtual machine based database hosting method under shared-nothing architecture. In this method, database requirement of tenant is satisfied by deploying database replica on the virtual machine. So the problem is how to optimize the resource (such as CPU, memory, etc.) allocation for the virtual machine which hosts database replica of tenant, to save resource cost while meeting the performance requirement of tenant. We model the above constraint programming problem, and solve the problem through a greedy algorithm based on the performance model and utility function. We concentrate on the impact on resource allocation of replica consistency, virtualization overhead and resource tuning granularity. The experiments show that, the algorithm can optimize more resource cost than other two representative resource allocation algorithms, while keeping the high performance in algorithm execution.(3) After realizing the importance of service management growing with the number and types of service hosted and operated in SaaS mode, we propose a service community model and corresponding derivation mechanism of service community, which are generalized from a real practice of SaaS. Especially, in order to assure service availability during service management, we also present an event based model for service monitoring and give its formalized definition. In the model, a service monitoring metamodel is put forward to define various service monitoring models on demand such that the monitored metrics, the monitor implementation and the monitoring process can be flexibly specified. Compared with the classic UDDI-based or semantic-oriented service management methods, service community have the features of flexibility and extendibility, and can enable the whole lifecycle management of services.Furthermore, based on above research results, we design and implement a SaaS mode information integration platform and form a systematic information integration solution and supporting software. The thesis also describes and discusses a real case study of applying the research results in the nationwide service network for sharing science and technology, which evaluates and shows the practicability of our research work. This work was supported by the National R&D Infrastructure and Facility Development Program of China (the grant No.2005DKA64201)...
|
PDF Full Text Request