SQL is not a perfect language for data mining, it’s by cunning tactics that Oracle got SQL accepted as standard, and Ingres lost. In fact if Ingres had won, it would have been much easier to query data today.
For all reasons bad SQL is here now and it’s here to stay, we can do almost nothing about it. While we interview Data Science candidates, we see that many are not proficient with SQL, and that’s a worry.
Since SQL had become a standard, lot of data is locked up and is still been pumped into and out of databases using SQL and no other way exists to get those data out for analytics. Many companies would prefer a data scientist who is good at SQL too so that they can cut out the cost of database administrator.
Some @ DataMaking strongly feel SQL should be phased out, but right now there is lot of data locked in Relational Database Management Systems (RDBMS), and almost all of them have SQL interface.
So if you want be a datascientist who would like to be hired, learn SQL. it would greatly help in getting a career or advancing it.
And if you ask what database to start with? We would suggest PostgreSQL as its top notch and most advanced RDBMS out there, and better its not controlled by any single corporation.