Apache Kylin : Analytical Data Warehouse for Big Data

Page tree

Welcome to Kylin Wiki.

Kylin 4.X recomend you to use RDBMS as metastore rather than HBase so you will not rely on HBase any more. But we did reserve HBase as secondary choice at the moment, but we may remove this support in near future.

HDP 2.X

  1. Uncomment the following code in bin/kylin.sh        
# uncomment the following code
if [ -n "${hbase_dependency}" ]; then
   hadoop_dependencies=${hadoop_dependencies}:${hbase_dependency}
fi

     2. Open conf/kylin.properties 

# add metadata url for HBase
kylin.metadata.url=kylin_metadata@hbase
# uncomment the following code
## uncomment for HDP
kylin.engine.spark-conf.spark.driver.extraJavaOptions=-Dhdp.version=current
kylin.engine.spark-conf.spark.yarn.am.extraJavaOptions=-Dhdp.version=current
kylin.engine.spark-conf.spark.executor.extraJavaOptions=-Dhdp.version=current
# add for hdp
kylin.query.spark-conf.spark.driver.extraJavaOptions=-Dhdp.version=current
kylin.query.spark-conf.spark.yarn.am.extraJavaOptions=-Dhdp.version=current
kylin.query.spark-conf.spark.executor.extraJavaOptions=-Dhdp.version=current

    3. start kylin

    bin/kylin.sh start

CDH 5.X

  1. Uncomment the following code in bin/kylin.sh 
kylin.sh
if [ -n "${hbase_dependency}" ]; then
   hadoop_dependencies=${hadoop_dependencies}:${hbase_dependency}
fi

    2. Edit $KYLIN_HOME/conf/kylin.properties

kylin.properties
# HBase Metadata url
kylin.metadata.url=kylin_metadata@hbase
# Connect to a remote zookeeper with the url, should set kylin.env.zookeeper-is-local to false
kylin.env.zookeeper-connect-string=hostname:port
  • No labels