[
https://issues.apache.org/jira/browse/HBASE-12394?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14194489#comment-14194489
]
Weichen Ye commented on HBASE-12394:
------------------------------------
I update some lineLengths issues and upload the patch.v2
Welcome to the ReviewBoard
https://reviews.apache.org/r/27519/
This patch is large, because I add three new classes and there are too much
redundant code between the following pairs:
TableMultiRegionInputFormat.java -- TableInputFormat.java
TableMultiRegionInputFormatBase.java -- TableInputFormatBase.java
TableMultiRegionMapReduceUtil.java -- TableMapReduceUtil.java
I do this because in this patch I hope nothing changes in the original code
TableInputFormat/TableInputFormatBase/TableMapReduceUtil.java. These 3 files
are the core of HBase-Mapreduce module. I hope the first patch only mind its
own business and do not affect others.
Now I am testing a new patch, which combine the redundant code together. In the
new patch all changes are merged into
TableInputFormatBase/TableInputFormat/TableMapReduceUtil.java, so the new patch
will be smaller. If there are no compatibility problems, I will upload the
smaller patch in few days.
So, for this patch , the only important part is the method getSplits(JobContext
context) in TableInputFormatBase.java.
The changes in other files are all about some references and class names, like
(TableInputFormat.SCAN ->TableMultiRegionInputFormat.SCAN).
> Support multiple regions as input to each mapper in map/reduce jobs
> -------------------------------------------------------------------
>
> Key: HBASE-12394
> URL: https://issues.apache.org/jira/browse/HBASE-12394
> Project: HBase
> Issue Type: Improvement
> Components: mapreduce
> Affects Versions: 2.0.0, 0.98.6.1
> Reporter: Weichen Ye
> Attachments: HBASE-12394.patch
>
>
> For Hadoop cluster, a job with large HBase table as input always consumes a
> large amount of computing resources. For example, we need to create a job
> with 1000 mappers to scan a table with 1000 regions. This patch is to support
> one mapper using multiple regions as input.
>
> The following new files are included in this patch:
> TableMultiRegionInputFormat.java
> TableMultiRegionInputFormatBase.java
> TableMultiRegionMapReduceUtil.java
> *TestTableMultiRegionInputFormatScan1.java
> *TestTableMultiRegionInputFormatScan2.java
> *TestTableMultiRegionInputFormatScanBase.java
> *TestTableMultiRegionMapReduceUtil.java
>
> The files start with * are tests.
> In order to support multiple regions for one mapper, we need a new property
> in configuration--"hbase.mapreduce.scan.regionspermapper"
> hbase.mapreduce.scan.regionspermapper controls how many regions used as input
> for one mapper. For example,if we have an HBase table with 300 regions, and
> we set hbase.mapreduce.scan.regionspermapper = 3. Then we run a job to scan
> the table, the job will use only 300/3=100 mappers.
> In this way, we can control the number of mappers using the following formula.
> Number of Mappers = (Total region numbers) /
> hbase.mapreduce.scan.regionspermapper
> This is an example of the configuration.
> <property>
> <name>hbase.mapreduce.scan.regionspermapper</name>
> <value>3</value>
> </property>
> This is an example for Java code:
> TableMultiRegionMapReduceUtil.initTableMapperJob(tablename, scan, Map.class,
> Text.class, Text.class, job);
>
>
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
