一種改進(jìn)的數(shù)據(jù)庫Sharding方法
發(fā)布時間:2018-09-13 05:51
【摘要】:Sharding是用于開源數(shù)據(jù)庫的一種水平擴展解決方案,包含兩個主要步驟:邏輯劃分和物理映射.對于前者,現(xiàn)有機制通常根據(jù)Shard Key采取基于Range或是Hash的方式來實現(xiàn).然而,基于Range的方法不能夠保證數(shù)據(jù)值域的均衡;基于Hash的方法不適用于范圍查詢,且當(dāng)Shard Key不均勻時無法保證數(shù)據(jù)量的均衡.對于后者,現(xiàn)有機制通常采用一致性哈希來確保存儲節(jié)點的動態(tài)擴展,然而該方法并未考慮數(shù)據(jù)增減導(dǎo)致Chunk大小變化時的數(shù)據(jù)均衡機制.為此,提出一種新的負(fù)載均衡的Sharding方法:一方面綜合考慮上述兩種劃分方式的優(yōu)缺點提出一種新的邏輯劃分方法,另一方面給出一種Chunk大小變化而導(dǎo)致的不均衡處理機制.實驗證明本文提出的方法能夠有效保證數(shù)據(jù)的負(fù)載均衡.
[Abstract]:Sharding is a horizontal extension solution for open source databases, which consists of two main steps: logical partitioning and physical mapping. For the former, the existing mechanism is usually based on Range or Hash based on Shard Key. However, the method based on Range can not guarantee the equilibrium of data range, and the method based on Hash is not suitable for range query, and can not guarantee the balance of data when Shard Key is not uniform. For the latter, the existing mechanism usually uses consistent hash to ensure the dynamic expansion of storage nodes. However, the method does not consider the data equalization mechanism when the data increase or decrease results in the change of Chunk size. Therefore, a new load balancing Sharding method is proposed. On the one hand, considering the advantages and disadvantages of the above two partitioning methods, a new logic partition method is proposed. On the other hand, a mechanism for dealing with the imbalance caused by the change of Chunk size is given. Experiments show that the proposed method can effectively ensure the load balance of the data.
【作者單位】: 東北大學(xué)計算機科學(xué)與工程學(xué)院;東北大學(xué)軟件學(xué)院;
【基金】:國家自然科學(xué)基金項目(612020888)資助 中央高;究蒲袠I(yè)務(wù)費專項基金項目(100704001)資助 教育部高等學(xué)校博士學(xué)科點專項科研基金項目(20120042110028)資助
【分類號】:TP311.13
本文編號:2240269
[Abstract]:Sharding is a horizontal extension solution for open source databases, which consists of two main steps: logical partitioning and physical mapping. For the former, the existing mechanism is usually based on Range or Hash based on Shard Key. However, the method based on Range can not guarantee the equilibrium of data range, and the method based on Hash is not suitable for range query, and can not guarantee the balance of data when Shard Key is not uniform. For the latter, the existing mechanism usually uses consistent hash to ensure the dynamic expansion of storage nodes. However, the method does not consider the data equalization mechanism when the data increase or decrease results in the change of Chunk size. Therefore, a new load balancing Sharding method is proposed. On the one hand, considering the advantages and disadvantages of the above two partitioning methods, a new logic partition method is proposed. On the other hand, a mechanism for dealing with the imbalance caused by the change of Chunk size is given. Experiments show that the proposed method can effectively ensure the load balance of the data.
【作者單位】: 東北大學(xué)計算機科學(xué)與工程學(xué)院;東北大學(xué)軟件學(xué)院;
【基金】:國家自然科學(xué)基金項目(612020888)資助 中央高;究蒲袠I(yè)務(wù)費專項基金項目(100704001)資助 教育部高等學(xué)校博士學(xué)科點專項科研基金項目(20120042110028)資助
【分類號】:TP311.13
【相似文獻(xiàn)】
相關(guān)期刊論文 前2條
1 呂學(xué)強,陳文亮,姚天順;基于連接文法的雙語E-Chunk獲取方法[J];東北大學(xué)學(xué)報;2002年09期
2 許廣永;薛靜;張建軍;;企業(yè)網(wǎng)絡(luò)運行與建構(gòu)中的動態(tài)均衡[J];長春工業(yè)大學(xué)學(xué)報(社會科學(xué)版);2012年04期
相關(guān)會議論文 前2條
1 劉昱;陳紅;王珊;;基于Chunk的緩存優(yōu)化與管理[A];第二十一屆中國數(shù)據(jù)庫學(xué)術(shù)會議論文集(技術(shù)報告篇)[C];2004年
2 盧艷民;焦有章;陳紅;;基于語義Chunk的動態(tài)實體化視圖技術(shù)[A];第二十二屆中國數(shù)據(jù)庫學(xué)術(shù)會議論文集(研究報告篇)[C];2005年
相關(guān)碩士學(xué)位論文 前2條
1 張振明;基于熱發(fā)布與動態(tài)平滑擴容的機票代購系統(tǒng)的設(shè)計與實現(xiàn)[D];哈爾濱工業(yè)大學(xué);2016年
2 吳永行;財經(jīng)問答系統(tǒng)問句解析預(yù)處理子系統(tǒng)的設(shè)計與實現(xiàn)[D];哈爾濱工業(yè)大學(xué);2015年
,本文編號:2240269
本文鏈接:http://sikaile.net/kejilunwen/ruanjiangongchenglunwen/2240269.html
最近更新
教材專著