Databases are converging, making it increasingly challenging to pinpoint the best use scenarios for each without understanding their underlying mechanics. This text explores Postgres, particularly its JSONB support, and compares it with NoSQL databases like Couchbase. Although Postgres is traditionally a relational database, its JSONB feature allows for JSON storage, offering advantages over traditional table structures, yet it is sometimes mistakenly seen as a NoSQL replacement. The article delves into the differences in data modeling between RDBMS and document databases, highlighting the complexity of JOIN operations in relational models compared to simpler, single-structure storage in NoSQL. It discusses the query languages of Postgres and Couchbase, noting that Postgres's JSONB is complex and lacks intuitive query capabilities, whereas Couchbase's N1QL is more SQL-like and user-friendly. Indexing in both systems is compared, with Couchbase offering more mature and flexible indexing options. The performance of Postgres's JSONB is promising but has limitations in dynamic models, while Couchbase excels in scalability and performance for specific use cases. The article concludes that while Postgres JSONB is valuable for certain transactional and metadata scenarios, Couchbase is better suited for applications requiring scalable performance and less rigid transactional support.