Improving HBase BulkLoad Performance
Scenario
BulkLoad uses MapReduce jobs to directly generate files that comply with the internal data format of HBase, and then loads the generated StoreFiles to a running cluster. Compared with HBase APIs, BulkLoad saves more CPU and network resources.
ImportTSV is an HBase table data loading tool.
This section applies to MRS 3.x and later versions.
Prerequisites
When using BulkLoad, the output path of the file has been specified using the Dimporttsv.bulk.output parameter.
Procedure
Add the following parameter to the BulkLoad command when performing a batch loading task:
|
Parameter |
Description |
Value |
|---|---|---|
|
-Dimporttsv.mapper.class |
The construction of key-value pairs is moved from the user-defined mapper to reducer to improve performance. The mapper only needs to send the original text in each row to the reducer. The reducer parses the record in each row and creates a key-value) pair.
NOTE:
When this parameter is set to org.apache.hadoop.hbase.mapreduce.TsvImporterByteMapper, this parameter is used only when the batch loading command without the HBASE_CELL_VISIBILITY OR HBASE_CELL_TTL option is executed. The org.apache.hadoop.hbase.mapreduce.TsvImporterByteMapper provides better performance. |
org.apache.hadoop.hbase.mapreduce.TsvImporterByteMapper and org.apache.hadoop.hbase.mapreduce.TsvImporterTextMapper |
Feedback
Was this page helpful?
Provide feedbackThank you very much for your feedback. We will continue working to improve the documentation.See the reply and handling status in My Cloud VOC.
For any further questions, feel free to contact us through the chatbot.
Chatbot
