; Hive concurrency lock
>
>
>
>
> wenye...@163.com
>
> 发件人： yu feng
> 发送时间： 2016-01-08 13:21
> 收件人： dev
> 主题： Re: How to improve the performance of job！
> According to our experience: you can try those :
> 1、use newer hive to promote the fir
in_job_conf.xml：
mapreduce.input.fileinputformat.split.maxsize
64MB
Hive concurrency lock
wenye...@163.com
发件人： yu feng
发送时间： 2016-01-08 13:21
收件人： dev
主题： Re: How to improve the performance of job！
According to our experience: you can try those :
1、use newer hive to promote the first
According to our experience: you can try those :
1、use newer hive to promote the first step.
2、startup more mapper and reducer for every MR job, you can reduce the
value of 'kylin.job.mapreduce.default.reduce.input.mb' in kylin.properties
which means input size for every reducer in NDCuboid calcula
I have five machines (8 core, 32g MEM), using HDP 2.3 building cluster
environment, version of the kyling Kyline
apache-kylin-1.3-HBase-1.1-SNAPSHOT-bin, HBase for Version 1.1.1, hive table
data is now 3000 ，but now job running the one hour, job schedule is about
10%, view the task of MR fo
