Modeling data and the approaches

The common way of data modeling for relational database is to reduce data redundancy and improve data integrity by using a normalization process. However, Elasticsearch is not ;a relational database and, as such, data modeling does not apply to it. Like most NoSQL databases, Elasticsearch treats the real world as a flat world, which means each document is independent. Basically, a single document should include all of the information used to determine whether it matches a search request. To bridge the gap between two worlds, several common techniques of data modeling are used in Elasticsearch. We'll discuss data denormalization, inner objects, nested objects, and join datatypes in the following sub-sections. No matter what kinds of technique you use, if you don't know the query that the user is running, you can't make good data modeling...