Building on Multi-Model Databases

Chapter 3. A Multi-Model Database for Data Integration

So, you’ve decided to use a multi-model database for data integration. What can you expect? How should it live in your infrastructure? Well, let’s take a look.

Your data has already been modeled. Very smart people in your organization modeled it to meet the requirements of various applications and business users. The data is just in silos. To integrate that data, you don’t want to remodel all the data to fit some new model. You want to take advantage of what you’ve already modeled as is and integrate it quickly to deliver results in a unified view.

Source systems for your integration might not go away. The data integration project is not necessarily a rip-and-replace solution. Some systems exist for very good reasons and will continue to exist; we just can’t use their presently contained data in isolation to achieve meaningful business results. So, if we need to integrate a silo’s data, a multi-model database can provide us an operational data hub and/or act as the glue for unifying various sources of data into a unified and consolidated layer. Let’s see how this plays out by stepping through some simple examples.

Entities and Relationships

Suppose that we have silos of customers, transactions, and products that we’d like to integrate into a unified view. Conceptually, the entities we care about and their relationships might look like Figure 3-1.

Figure 3-1. Entities and relationships

A customer can have many transactions; ...