Database Reference
In-Depth Information
NOTE
In a default Hive setup, the Derby database used for the metastore may
be configured for single-user access. If you are just testing Hive or
running a local instance for development, this may be fine. However,
for Hive implementation in a production environment, you will want to
upgrade the metastore to a multiple-user setup using a more robust
database. One of the more common databases used for this is MySQL.
However, the metastore can be any Java Database Connectivity
(JDBC)-compliant database. If you are using the Hortonworks' HDP 1.3
Windows distribution, SQL Server can be used as a supported
metastore.
Hive v0.11 also includes HiveServer2. This version of Hive improves
support for multi-user concurrency and supports additional
authentication methods, while providing the same experience as the
standard Hive server. Again, for a production environment,
HiveServer2 may be a better fit. The examples used in this chapter run
against Hive Server and HiveServer2.
Another area of difference between Hive and many relational databases is
itssupportfordifferentdatatypes.Duetotheunstructureddatathatitmust
support, itdefines anumber ofdata typesthatyouwon'tfind inatraditional
relational database.
Hive Data Types
Table 6.1
lists the data types supported by Hive. Many of these data types
have equivalent values in SQL Server, but a few are unique to Hive. Even for
the data types that appear familiar, it is important to remember that Hive
is coded as a Java application, and so these data types are implemented in
Java. Their behavior will match the behavior from a Java application that
uses the same data type. One immediate difference you will notice is that
STRING
types do not have a defined length. This is normal for Java and
other programming languages, but is not typical for relational databases.
