一种基于中间件的异构数据库融合访问方法及系统 (Design and Implemention of Accessing Hybrid Database Systems Based on Middleware)

Mingming Pan, Dingding Li, Yong Tang, Hai Liu

Abstract: In the big data era,with the explosive growth of informational data,traditional relational databases(SQL) and emerging NoSQL databases are difficult to face these challenges in a comprehensive and efficient manner.Therefore,this paper proposed MingleDB,a high-efficient middleware for incorporating both merits of traditional database and NoSQL systems.MingleDB is transparent for the underlying hybrid database systems,namely SQL and NoSQL,without any intrusive modifications on the original systems.MingleDB can detect the specific characteristic inside a user query,for example,the data is either unstructured or structured.Then it puts this user query into the suitable procedure(SQL or NoSQL) presented by MingleDB,and also provides a lightweight transaction processing framework which is implemented on demand to ensure the final consistency and completeness of hybrid database data.Extensive experiments were conducted to test the effective of MingleDB.Furthermore,this paper used a realistic scenario to verify the advantage of our work.The results are positive.