[
https://issues.apache.org/jira/browse/PIG-3648?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13863730#comment-13863730
]
Aniket Mokashi edited comment on PIG-3648 at 1/7/14 12:51 AM:
--------------------------------------------------------------
We are using reservoir sampling here, with assumption that the number of
samples fit in memory. My only question/concern is how much benefit does
increasing sample size provide here.
In your example- 100 samples on 13M rows had 10x skew. Does 200 samples make it
5x skew or less? If it does, doing this definitely makes sense.
was (Author: aniket486):
We are using reservoir sampling here, with assumption that the number of
samples fit in memory. My only question/concern is how much benefit does
increasing sample size provide here.
In your example- 100 samples on 13M rows had 10x skew. Does 200 sample make it
5x skew or less? If it does, doing this definitely makes sense.
> Make the sample size for RandomSampleLoader configurable
> --------------------------------------------------------
>
> Key: PIG-3648
> URL: https://issues.apache.org/jira/browse/PIG-3648
> Project: Pig
> Issue Type: Improvement
> Components: impl
> Reporter: Cheolsoo Park
> Assignee: Cheolsoo Park
> Priority: Minor
> Fix For: 0.13.0
>
> Attachments: PIG-3648-1.patch
>
>
> Pig uses RandomSampleLoader for range partitioning in order-by. But since the
> sample size is hardcoded as 100, volatility in the variance of the results
> increases when sorting a large number of rows (e.g. 10M+ per task).
> It would be nice if the sample size could be configurable via Pig properties.
--
This message was sent by Atlassian JIRA
(v6.1.5#6160)
