[jira] [Updated] (HIVE-6473) Allow writing HFiles via HBaseStorageHandler table

["Nick Dimiduk (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/HIVE-6473?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Nick Dimiduk updated HIVE-6473:
-------------------------------

    Attachment: HIVE-6473.0.patch.txt

This patch introduces a new configuration flag hive.hbase.generatehfiles. When it is enabled, the Storage Handler will use HiveHFileOutputFormat for writing new records.

Note that all existing limitations for HFile generation from Hive remain, notably:
 - the property hfile.family.path must be specified either as a table property or via job configuration
 - only a single column family is supported
 - records must arrive at the RecordWriter in a sorted order that is compatible with HFileWriter requirements

> Allow writing HFiles via HBaseStorageHandler table
> --------------------------------------------------
>
>                 Key: HIVE-6473
>                 URL: https://issues.apache.org/jira/browse/HIVE-6473
>             Project: Hive
>          Issue Type: Improvement
>          Components: HBase Handler
>            Reporter: Nick Dimiduk
>         Attachments: HIVE-6473.0.patch.txt
>
>
> Generating HFiles for bulkload into HBase could be more convenient. Right now we require the user to register a new table with the appropriate output format. This patch allows the exact same functionality, but through an existing table managed by the HBaseStorageHandler.



--
This message was sent by Atlassian JIRA
(v6.1.5#6160)