Font Size: a A A

Research Of Data Allocation In Distributed Database

Posted on:2005-03-13Degree:MasterType:Thesis
Country:ChinaCandidate:Y YangFull Text:PDF
GTID:2168360125963933Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
The storage and application of Data become distributed increasingly with the development of computer network and distributed location of the corporations and organizations. Distributed database technology appeared based on that environments. That technology has more advantages than that in traditional centralized database in aspect of reliability, usability and time response, so it is applied widely in the real world.Data is not stored centrally but to be distributed in different sites according to application requirements, so data should be partitioned in distributed database system. That is fragmentation and fragment allocation issue should be concerned in distributed database system. Redundant data can increase local data processing, improve system efficiency, reliability and usability. One of goal of distributed database design is data processing locally, namely allocating fragments to the issuing sites so that the remote communication cost will be reduced.How to allocate fragments rationally must be considered in distributed database system. Fragment allocation issue has a significant impact on the system performance, usability, efficiency and reliability. System performance is well only if fragment allocation is well. Researchers both domestic and overseas have been doing research on fragment allocation in distributed database system, but a general fragment allocation method of combining optimal cost and performance has not been found for its complication. Although a large amount of researchers have proposed many models and algorithms designed to allocate fragments, most of their models and algorithms are very complicated and not well understood. Thus, it is difficult to use them in a real environment. Here, I propose a heuristic method called remove fragment copies allocation based on researching and modifying the Benefit-cost heuristic fragment allocation method. [16] In this paper, firstly, some key factors influenced how to select statistical information is concluded, then a communication cost formulas for transaction processing is derived, finally, a heuristic remove fragment copies allocation algorithm is proposed. The algorithm has two steps. In the first step, the fragment allocation table by only considering the retrieval requests issued by transactions is initialized without any communication cost by allocating its target fragments to its issuing site. In the second step, the issue that to be considered is to remove fragments copies from the initialized allocation such that the communication cost will be reduced for all requests. At last, the method is realized by experiment. It has been proved that it is simple, comprehensive, and can reduce the scope of seeking solution, the result can more reflect the real world.
Keywords/Search Tags:Distributed Database, Fragment Allocation, Statistical Information, Cost Formulas, Heuristic Allocation Algorithm
PDF Full Text Request
Related items