A few side notes:
- One of the historic rivals to SQL is [Datalog](https://en.wikipedia.org/wiki/Datalog). SQL was released in 1974 and researchers started having workshops about datalog in 1977.
Datalog is a based on [Prolog](https://en.wikipedia.org/wiki/Prolog). Prolog is a programming language based on the logic of [Horn clauses](https://en.wikipedia.org/wiki/Horn_clause).
Datalog proponents used to critique SQL as not supporting recursive queries. However in [SQL 1999](https://en.wikipedia.org/wiki/SQL:1999) recursive queries were introduced. The two query languages are thought to have the same power now.
In industry, there are a few people still using Datalog. There's [Datomic](https://www.datomic.com/) which is a product by Cognitect, the company behind [clojure](https://en.wikipedia.org/wiki/Clojure). Datomic wraps postgres and conventional SQL solutions.
There is also [Cascalog](http://cascalog.org/). Cascalog uses apache Hadoop to drive queries over large data sets. It hasn't been in development for a couple of years, so I think the industry has moved on.
- Postgres in particular is strongly consistent as far as the CAP theorem is concerned. Jepsen has done some excellent stress testing of postgres in his [Jepsen blog](https://aphyr.com/posts/282-jepsen-postgres).
- Modern databases support a wide variety of data structures. One in particular that is useful for geographic information systems (GIS) is the [R-tree](https://en.wikipedia.org/wiki/R-tree). These have very good lookup characteristics for *nearest neighbor* searches. You can use R-trees in PostgresSQL and MySQL.