Kyuubi v.s. HiveServer2#


HiveServer2 is a service that enables clients to execute Hive QL queries on Hive supporting multi-client concurrency and authentication. Kyuubi enables clients to execute Spark SQL queries directly on Spark supporting multi-client concurrency and authentication too.

They are both designed to provide better support for open API clients like JDBC and ODBC to manage and analyze BigData.

Hive on Spark#

The purpose of Hive on Spark is to add Spark as a third execution backend, parallel to MR and Tez. Comparing to Hive on MR, it’s use the Spark DAG will help improve the performance of Hive queries, especially those have multiple reducer stages.

Differences Between Kyuubi and HiveServer2#

Kyuubi HiveServer2
Language Spark SQL Hive QL
Optimizer Spark SQL Catalyst Hive Optimizer
Engine up to Spark 3.x MapReduce/up to Spark 2.3/Tez
Performance High Low
Compatibility with Spark Good Bad(need to rebuild on a specific version)
Data Types Spark Data Types Hive Data Types



  1. HiveServer2 Overview