Hundreds of thousands of small servers create distributed computing instead of a single powerful machine.Data needs to be transformed to a different model to support distributed computing.
So data needs to be distributed, and data schema needs to be de-normalized according to the business requirements.
Schemas should be designed for enabling distributed query.
This requires each data set to contain enough information to run the executed queries separately in different data nodes.
In this article, I discuss a solution for managing both No SQL and relational databases using the Unified Data Modeling techniques.
Unified data modeling supports features like document schema of No SQL databases and reverse engineering of data from an existing database.
It also supports visual refactoring of existing databases.With data growth on 4V’s (Volume, Variety, Velocity, Value), data management is evolving from scaling up to scaling out.Nowadays, No SQL databases co-exist with relational databases in enterprise data architecture.However, No SQL data management currently lacks mature methods and tools to manage No SQL data as well as relational data.Most existing No SQL databases are designed with more consideration on application performance, less on high level business models, data integration and data standardization.There is a gap between data modeling and physical data aspects of No SQL databases.