hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Vasu Mariyala (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-10416) Improvements to the import flow
Date Tue, 28 Jan 2014 19:28:38 GMT

     [ https://issues.apache.org/jira/browse/HBASE-10416?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel

Vasu Mariyala updated HBASE-10416:

    Release Note: 
Import with this fix supports 

a) Filtering of the row using the Filter#filterRowKey(byte[] buffer, int offset, int length).

b) Accepts durability parameter (Ex: -Dimport.wal.durability=SKIP_WAL ) while importing the
data into HBase. If the data doesn't need to be replicated to the DR cluster or if the same
import job would be run on the dr cluster, consider using SKIP_WAL durability for performance.

> Improvements to the import flow
> -------------------------------
>                 Key: HBASE-10416
>                 URL: https://issues.apache.org/jira/browse/HBASE-10416
>             Project: HBase
>          Issue Type: New Feature
>          Components: mapreduce
>            Reporter: Vasu Mariyala
>         Attachments: HBASE-10416.patch
> Following improvements can be made to the Import logic
> a) Make the import extensible (i.e., remove the filter from being a static member of
Import and make it an instance variable of the mapper, make the mappers or variables of interest
protected. )
> b) Make sure that the Import calls filterRowKey method of the filter (Useful if we want
to filter the data of an organization based on the row key or using filters like PrefixFilter
which filter the data in filterRowKey method rather than the filterKeyValue method). The existing
test case in TestImportExport#testWithFilter works with this assumption but is so far successful
because there is only one row inserted into the table.
> c) Provide an option to specify the durability during the import (Specifying the Durability
as SKIP_WAL would improve the performance of restore considerably.) [~lhofhansl] suggested
that this should be a parameter to the import.
> d) Some minor refactoring to avoid building a comma separated string for the filter args.

This message was sent by Atlassian JIRA

View raw message