| Summary: | 碩士 === 國立高雄應用科技大學 === 資訊管理研究所碩士班 === 102 === With the growing popularity of big data and cloud computing, the loading on traditional relational database is heavy. To avoid affecting others applications in an enterprise, the common approach is to add additional databases to reduce the loading of the original databases.
In 1998, Carlo Strozzi proposed the concept of NoSQL, in which there is no correlation between data records. Unlike relational databases, NoSQL does not need a predefined schema, and the expansion of data storage in NoSQL is easy, that is, when a user needs storage expansion, he simply adds server nodes. Thus, the users can increase or decrease the required storage space in accordance with their own needs.
A relational database needs normalization in defining a data model (i.e., schema), which helps developers confirm the correctness of the defined data model. Relational database can effectively control the integrity and aggregation of the queried data by using SQL(Structured Query Language).
NoSQL database is an aggregate-oriented data model. During the design phase, developers must know what is the needed information for the system, and then they design the data model according to the needed information.
In traditional relational databases, developers define the schema of a database and then use standard SQL commands to query data. In NoSQL database, developers first develop a data model, which may varies for different developers due to their different perspectives, and then uses the data model-specific commands to query data.
It is still not clear how to define a well-performed data model for a NoSQL database, especially for query commands. In view of this, this thesis investigates how to design a well-performed data model for NoSQL databases.
Different data models may have different data computing efficiency and require different storage capacity. When there exists a complex relationship between the data sets in a NoSQL database, it is important to explore an efficient data model to improve the overall execution performance.
|
|---|
