[
https://issues.apache.org/jira/browse/PHOENIX-2786?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15204754#comment-15204754
]
maghamravikiran commented on PHOENIX-2786:
------------------------------------------
[~churromorales] From what I see, MultiTableOutputFormat uses the Put / Delete
mutation rather than writing to HFiles that MultiHfileOutputFormat does. We
definitely have seen times , for ex: for a newly created table , doing direct
writes to HBase perform way better than bulk load but in general writing to
HFiles performs better.
I definitely agree to your valid point that the code in MultiHfileOutputFormat
has a lot from HfileOutputFormat except for few minor changes.
> Can MultiTableOutputFormat be used instead of MultiHfileOutputFormat
> --------------------------------------------------------------------
>
> Key: PHOENIX-2786
> URL: https://issues.apache.org/jira/browse/PHOENIX-2786
> Project: Phoenix
> Issue Type: Task
> Reporter: churro morales
>
> MultiHfileOutputFormat depends on a lot of HBase classes that it shouldn't
> depend on. It seems like MultiHfileOutputFormat and MultiTableOutputFormat
> have the same goal.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
