[
https://issues.apache.org/jira/browse/PARQUET-225?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15742264#comment-15742264
]
Jakub Liska commented on PARQUET-225:
-------------------------------------
Guys how do I enable it? I tried to encode :
{code:java}
Types.required(PrimitiveTypeName.INT64).named("foo")
{code}
that starts from 0 being incremented by 1, but I the resulting parquet file
size is identical in 1.9.0 and 1.8.0 versions...
Is there something I have to enable for Delta Encoding to get applied?
> INT64 support for Delta Encoding
> --------------------------------
>
> Key: PARQUET-225
> URL: https://issues.apache.org/jira/browse/PARQUET-225
> Project: Parquet
> Issue Type: New Feature
> Reporter: Vassil Lunchev
> Assignee: Vassil Lunchev
> Fix For: 1.9.0
>
>
> As of now, parquet doesn't support delta encoding for INT64. However it is
> planned in the format:
> https://github.com/Parquet/parquet-format/blob/master/Encodings.md
> The benefits of this feature are huge. For timestamps it achieves twice
> better compression than SNAPPY on plain encoding, and the reading is faster.
> This feature is actually advertised on the home page of Parquet, even though
> it is not yet implemented:
> http://parquet.incubator.apache.org/
> http://image.slidesharecdn.com/hadoopsummit-140630160016-phpapp01/95/efficient-data-storage-for-analytics-with-apache-parquet-20-30-638.jpg?cb=1404162126
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
