Database schema design has a significant importance in software design. There are lots of tools and methods available for schema design in RDBMS but limited attention is given in NoSQL for schema design as it is emerging in database technology. NoSQL requires a different approach in designing efficient schema, such as in the document database which information should be stored as embedded document, or which information should be stored as referenced document. There are certain thumb rules for schema design in NoSQL databases. In reengineering projects, especially in Old RDBMS to new NoSQL system, developing good and efficient database schema is a very difficult task. In this paper, we have proposed a schema design advisor model which uses the existing software’s SQL queries load as an input along with an algorithm for schema design recommendation. Also, we have proposed a cost model for various schemas created by recommendation model. The proposed model is implemented through a prototype for the MongoDB document database in Java. The prototype produces all possible combinations of schemas and calculates cost of each schema. Automated schema design process produces all possible combinations of NoSQL schemas, which is difficult with manual schema design approach.

数据库模式设计在软件设计中非常重要。RDBMS中有许多工具和方法可用于模式设计，但是NoSQL在模式设计中的关注很少，因为它在数据库技术中不断涌现。NoSQL在设计高效模式时需要不同的方法，例如在文档数据库中，哪些信息应存储为嵌入式文档，或者哪些信息应存储为引用文档。NoSQL数据库中有一些用于架构设计的经验法则。在重新设计项目时，尤其是从旧的RDBMS到新的NoSQL系统，开发良好而高效的数据库架构是一项非常困难的任务。在本文中，我们提出了一个架构设计顾问模型，该模型使用现有软件的SQL查询负载作为输入以及架构设计推荐算法。此外，我们还为推荐模型创建的各种模式提出了成本模型。所提出的模型是通过Java中的MongoDB文档数据库的原型实现的。原型产生所有可能的模式组合，并计算每个模式的成本。自动化的方案设计过程会生成NoSQL方案的所有可能组合，而手动方案设计方法很难做到这一点。