Bigdata SQL: Apache Hive Architecture Deep Dive

Hive uses a relational store to store the metadata for all the database objects it manages. This relational store is typically either MySQL or Postgres, and it contains the metadata
for all the objects: databases, database properties, tables, views, table column data types, file formats for the table, file locations, partitioning and bucketing details, etc. The
metastore has a Thrift interface, whereby it can be accessed by clients written in different programming languages.