NoSQL -> programmer productivity, data access performance.

Relational DBs don't work well with data that has many relationships.

Polyglot persistence -> uses multiple data storage solutions.

2000 popular frameworks like: Hibernate, iBATIS and Rails Active Record.

In the 1990s many projects failed in the effort to build object-relational mapping layers.

The best possible evaluation should be made, knowing the defects.

Working a week using technology is much more efficient than watching and learning hundreds of presentations.

The need that drove DB NoSQL -> fast access to big data.

Aggregate-oriented DBs can be very fast at reading or retrieving aggregates.

Fragmentation and replication allow horizontal scaling.

Properly test -> create something, run it and measure it.

Better to stick with the default option of a relational DB.

When selecting a NoSQL DB, you must demonstrate concrete advantages over relational databases in your context.

To change the choice the most classic way to do it is through a data warehouse layer like Data Mapper and Repository.

By wrapping sections of your DB in services, you can replace parts with NoSQL technology as the advantages become clear over time.

Reasons to use NoSQL technology -> better productivity and improve performance.

Encapsulation supports changes in storage technologies.