【摘要】：隨著移動互聯(lián)網(wǎng)的快速發(fā)展,基于位置服務(wù)的電子商務(wù)受到越來越多的人的歡迎,但是電子商務(wù)平臺的數(shù)據(jù)量龐大,傳統(tǒng)的集中式數(shù)據(jù)庫不能很好地解決數(shù)據(jù)存儲。本文以解決電子商品平臺海量數(shù)據(jù)存儲為目的,設(shè)計實現(xiàn)了基于中間件的分布式數(shù)據(jù)庫存儲方案。由于電子商務(wù)系統(tǒng)業(yè)務(wù)復(fù)雜、數(shù)據(jù)庫的一致性要求較高,因此采用基于關(guān)系型的分布式數(shù)據(jù)庫存儲是最合適的選擇。在本文設(shè)計的分布式數(shù)據(jù)庫基礎(chǔ)上,本文還實現(xiàn)了由該分布式數(shù)據(jù)庫提供支持的基于位置服務(wù)的電子商務(wù)系統(tǒng)。論文主要工作如下:(1)對分布式數(shù)據(jù)庫存儲技術(shù)進行了研究,分析了各個存儲方案的優(yōu)缺點。以分布式數(shù)據(jù)庫分片算法為重點,分析了各種分片算法存在的不足。然后詳細分析了谷歌提出的高性能跳增一致性哈希算法,并結(jié)合分布式文件系統(tǒng)Ceph內(nèi)部的CRUSH算法提出了一種考慮節(jié)點權(quán)重的分片算法——加權(quán)跳增一致性哈希分片算法。最后把該算法運用到數(shù)據(jù)分片中。該算法以實現(xiàn)數(shù)據(jù)均勻分布為目的,以考慮節(jié)點權(quán)重為重點,充分滿足了對分片算法性能、平衡性和單調(diào)性的要求。(2)詳細研究并分析了分布式數(shù)據(jù)庫中的分片連接方案,指出了在普通連接操作下進行分片連接的高成本問題。為了降低分布式數(shù)據(jù)庫集群環(huán)境中分片連接的通信成本,提出了一個基于布隆過濾器的分片連接方案。本文對該方案進行了詳細的闡述和分析,并通過數(shù)學(xué)計算和函數(shù)模型得出該方案的可行性。(3)在分片算法的基礎(chǔ)之上,并結(jié)合分布式數(shù)據(jù)庫中間件Mycat,實現(xiàn)了分布式數(shù)據(jù)庫存儲方案。使用開源工具Otter完成了對數(shù)據(jù)庫的同步備份。介紹了系統(tǒng)的需求分析、架構(gòu)設(shè)計以及系統(tǒng)關(guān)鍵模塊的實現(xiàn),最終設(shè)計并實現(xiàn)了的基于地理位置服務(wù)的分布式電子商務(wù)系統(tǒng)。在本文最后通過大量實驗驗證了本文提出的分片算法的性能、平衡性和單調(diào)性。得出該分片算法在性能上較好,數(shù)據(jù)分布也比較均勻,同時在節(jié)點權(quán)重相等的條件下單調(diào)性較好,但如果改變靠近首節(jié)點的權(quán)重則會引起大規(guī)模的數(shù)據(jù)遷移。
[Abstract]:With the rapid development of mobile Internet, e-commerce based on location-based services is welcomed by more and more people. However, the amount of data on e-commerce platform is huge, and the traditional centralized database can not solve the problem of data storage very well. In order to solve the problem of mass data storage in electronic commodity platform, this paper designs and implements a distributed database storage scheme based on middleware. Because of the complexity of e-commerce system and the high requirement of database consistency, it is the most appropriate choice to use distributed database storage based on relationship. On the basis of the distributed database designed in this paper, a location-based e-business system supported by the distributed database is implemented. The main work of this paper is as follows: (1) the distributed database storage technology is studied, and the advantages and disadvantages of each storage scheme are analyzed. Based on the distributed database slicing algorithm, the shortcomings of each algorithm are analyzed. Then, the high performance jump consistency hash algorithm proposed by Google is analyzed in detail. Combined with the CRUSH algorithm within the distributed file system Ceph, a weighted jump consistent hash algorithm, which considers the node weight, is proposed. Finally, the algorithm is applied to data slicing. In order to achieve uniform distribution of data, this algorithm fully satisfies the requirements of the performance, balance and monotonicity of the fragmentation algorithm. (2) the piecewise connection scheme in distributed database is studied and analyzed in detail, with the emphasis on considering node weight, and the algorithm meets the requirements of the performance, balance and monotonicity of the algorithm. The high cost of piecewise connection under common connection operation is pointed out. In order to reduce the communication cost of piecewise connection in distributed database cluster environment, a piecewise connection scheme based on Bloom filter is proposed. In this paper, the scheme is elaborated and analyzed in detail, and the feasibility of the scheme is obtained by mathematical calculation and function model. (3) based on the slicing algorithm and combined with the distributed database middleware Mycat, the feasibility of the scheme is obtained. The distributed database storage scheme is implemented. Using the open source tool Otter to complete the synchronous backup of the database. This paper introduces the requirement analysis, architecture design and the implementation of the key modules of the system. Finally, the distributed electronic commerce system based on geographical location service is designed and implemented. At the end of this paper, the performance, balance and monotonicity of the proposed piecewise algorithm are verified by a large number of experiments. It is concluded that the proposed algorithm has good performance and uniform data distribution. At the same time, the monotonicity is better under the condition of equal node weight, but if the weight near the first node is changed, the large-scale data migration will be caused.
【學(xué)位授予單位】：電子科技大學(xué)
【學(xué)位級別】：碩士
【學(xué)位授予年份】：2017
【分類號】：F713.36;TP311.13

【參考文獻】

相關(guān)期刊論文 前3條

1 申德榮;于戈;王習(xí)特;聶鐵錚;寇月;;支持大數(shù)據(jù)管理的NoSQL系統(tǒng)研究綜述[J];軟件學(xué)報;2013年08期

2 王利娟;馬曉;;基于LBS的移動電子商務(wù)商業(yè)模式探析——LBS前途光明,道路曲折[J];企業(yè)導(dǎo)報;2012年02期

3 孟小峰,周龍驤,王珊;數(shù)據(jù)庫技術(shù)發(fā)展趨勢[J];軟件學(xué)報;2004年12期

，

本文編號：2430241