I know the answer to this question largely depends on the use case and specific business scenario but for folks in consulting roles, based on your experience what is becoming the new norm: What are the majority of the companies doing these days in terms of normalizing vs de normalizing data? Is the industry moving away from p-f keys to more more open source and no Sql DBs? How is the general/ broad/high level data layout?
Everyone is moving away from relational databases and it's because everyone is idiotic. It's a hype wave that started about 10 years ago and got REALLY intense around 2012–2014. It's bullshit. Absolute uneducated vacuous ignorant ridiculous clown fucking shit. Generally, the bigger the company, the less it understands data and how to architect and evolve a data-intensive application. In any case, the licensing and source distribution model is absolutely orthogonal to the data model.
This. Nosql has some use cases but for the majority of data relational databases are going to be the king for a very long time
Totally agree. Lazy app devs want to throw everything in big json blob in mongodb or dyanamo. When time comes to consume by DW team..they pull their hair what to make of it or create complicated ETL and try to map to relational DB.
Open source vs. relational is a false dichotomy: MySQL, Postgres are two popular open-source relational databases. You should also consider databases-as-a-service such as CloudSQL, Spanner, Aurora.
Agreed. Even in an “as a service” model I was wondering what the data layout approach is these days? Parent child relation established or more towards redundant data and faster reads as opposed to p-f keys?