【HIVE & Spark】将hive引擎换成Spark，运行速度快！怎么换？请看本文

本教程仅仅是使用spark,能在hive用就行。

1.下载Spark;

2.WinSCP上传spark压缩包到虚拟机；

3.tar -zxvf spark-2.3.3-bin-without-hadoop -C /opt/programs/

4.

1）配置Spark环境变量；

2）配置spark-env.sh 命令cp spark-env.sh.template spark-env.sh后vi spark-env.sh

末尾加上：

 export JAVA_HOME=/usr/java/jdk1.8.0_25
    export SPARK_DIST_CLASSPATH=$(/opt/programs/hadoop-2.6.0/bin/hadoop classpath)

// /opt/programs/hadoop-2.6.0/bin/hadoopg改成自己的路径

3）配置slaves

hadoop1

hadoop2

hadoop3

4）将Spark相关文件，连接到$HIVE_HOME/lib中

3个包：

scala-library-2.11.8.jar

spark-core_2.11-2.3.3.jar

spark-network-common_2.11-2.3.3.jar

ln -snf /opt/programs/spark-2.3.3/jars/spark-core_2.11-2.3.3.jar   /opt/programs/hive-2.3.5/lib/spark-core_2.11-2.3.3.jar

配置成功的结果：

[root@hadoop1 conf]# ll /opt/programs/hive-2.3.5/lib/ | grep spark
		lrwxrwxrwx. 1 root root       55 Sep 12 22:26 scala-library-2.11.8.jar -> /opt/programs/spark-2.3.3/jars/scala-library-2.11.8.jar
		lrwxrwxrwx. 1 root root       56 Sep 12 22:27 spark-core_2.11-2.3.3.jar -> /opt/programs/spark-2.3.3/jars/spark-core_2.11-2.3.3.jar
		lrwxrwxrwx. 1 root root       66 Sep 12 22:27 spark-network-common_2.11-2.3.3.jar -> /opt/programs/spark-2.3.3/jars/spark-network-common_2.11-2.3.3.jar

5）配置hive执行引擎

1. 在配置文件里面配置；

 			<property>
                    <name>hive.execution.engine</name>
                    <value>spark</value>
    		</property>

1. 在beeline配置，只在当前session有效；

//在beeline里
set hive.execution.engine=spark;