hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Ted Yu (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (HBASE-16637) Add endpoint coprocessor to speed up queries to hbase:backup table
Date Tue, 20 Dec 2016 16:36:58 GMT

     [ https://issues.apache.org/jira/browse/HBASE-16637?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Ted Yu updated HBASE-16637:
---------------------------
    Description: 
Currently BackupSystemTable uses HTable API to query backup table.
In many scenarios, this is inefficient because filtering is done at client side.

With more information being recorded in hbase:backup table (such as bulk load descriptor -
see HBASE-14417), we should perform filtering on server side.

An endpoint coprocessor would serve this purpose.
e.g. without endpoint coprocessor, incremental restore would either bring all bulk load descriptors
to client followed by filtering for relevant tables or, issue two queries - one for rows related
to the tables being restored and another query to bring down the bulk load descriptors.

  was:
Currently BackupSystemTable uses HTable API to query hbase:backup table.
In many scenarios, this is inefficient because filtering is done at client side.

With more information being recorded in hbase:backup table (such as bulk load descriptor -
see HBASE-14417), we should perform filtering on server side.

An endpoint coprocessor would serve this purpose.
e.g. without endpoint coprocessor, incremental restore would either bring all bulk load descriptors
to client followed by filtering for relevant tables or, issue two queries - one for rows related
to the tables being restored and another query to bring down the bulk load descriptors.


> Add endpoint coprocessor to speed up queries to hbase:backup table
> ------------------------------------------------------------------
>
>                 Key: HBASE-16637
>                 URL: https://issues.apache.org/jira/browse/HBASE-16637
>             Project: HBase
>          Issue Type: Improvement
>            Reporter: Ted Yu
>              Labels: backup
>
> Currently BackupSystemTable uses HTable API to query backup table.
> In many scenarios, this is inefficient because filtering is done at client side.
> With more information being recorded in hbase:backup table (such as bulk load descriptor
- see HBASE-14417), we should perform filtering on server side.
> An endpoint coprocessor would serve this purpose.
> e.g. without endpoint coprocessor, incremental restore would either bring all bulk load
descriptors to client followed by filtering for relevant tables or, issue two queries - one
for rows related to the tables being restored and another query to bring down the bulk load
descriptors.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message