database.txt

NoSQL vs SQL:
    1. MongoDB is very popular. (Yu:  used by vnpy and quokka)
    2. The db size grow rapidly since nosql stores in json format and the keys are necessary for everything
    3. Transaction is critical at certain conditions.
    4. NoSQL is widely taken as esay to use. It is indeed but the syntax is not so clean.
    5. NoSQL is very good for unstructured data and horizontal scalability
    6. Above items are from web search. 

SQL: (We choose postgresql)
    1. Very stable over past 20 years and widely available in almost all OS
    2. Still in rapid developing
    3. Both native replication/clustering capability (hot-standby) and 3rd party solutions (Citus) exist

Time-Series DB:
    1. InfulxDB is solid choice. (Check grafana). Also check Prometheus
    2. Kafka and TimeScaleDB looks interesting too (Kafaka for superset is my next project, see my comment in the issue)
    3. Recently (2017/4/5) a new time-series DB (TimescalDB) is developed based on postgresql (Ref: "Building a scalable time-series database on postgreSQL")
    4. Tgres is another time-series db build on PostSQL written in Go, but without extension requirement. (Yu: sounds better?)