After years of NoSQL programming, people start to understand the importance, flexibility, power and spread of SQL as a recognized standardized language. More and more NoSQL databases start to offer SQL-like extensions (e.g. Couchbase N1QL). SQL-on-Hadoop is also very popular. Technologies like Oracle Big Data SQL or Microsoft Polybase retrieve data from heterogeneous databases through an SQL interface.
- How to aggregate key figures stored in thousands of JSON documents?
- How to scan through various sensor logs containing different formats and versions?
- How to detect data quality issues if there is no agreed model?
Schema-free design comes with hidden costs like structural complexity and lack of analytics. Iggy Fernandez addresses other issues in The Rise and Fall of the NoSQL Empire. NoSQL introduced many good ideas. But:
The revolution will not be schema-less 🙂
— Gwen (Chen) Shapira (@gwenshap) 3. Dezember 2015
SQL always was a strong language and SQL will remain a strong language. RDBMSes like Oracle offer even more powerful extensions like the model clause or pattern matching to identify data patterns without having to write long programs. More code means more errors. Only the Strong Survive (songtext).