[ 
https://issues.apache.org/jira/browse/BEAM-6243?focusedWorklogId=186656&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-186656
 ]

ASF GitHub Bot logged work on BEAM-6243:
----------------------------------------

                Author: ASF GitHub Bot
            Created on: 18/Jan/19 01:36
            Start Date: 18/Jan/19 01:36
    Worklog Time Spent: 10m 
      Work Description: robertwb commented on issue #7297:  [BEAM-6243] Add an 
experiment to use Python's optimizer on Flink.
URL: https://github.com/apache/beam/pull/7297#issuecomment-455395621
 
 
   Mostly Java is very conservative when it comes to fusing anything that has 
any side inputs. That doesn't really come up here though; the code in question 
her is all about taking the list of stages from the Python optimizer (whatever 
decisions it made) and re-constructing a pipeline proto with executable stage 
transforms. 
 
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org


Issue Time Tracking
-------------------

    Worklog Id:     (was: 186656)
    Time Spent: 40m  (was: 0.5h)

> TFX pipelines experience a huge blowup in intermediate data size
> ----------------------------------------------------------------
>
>                 Key: BEAM-6243
>                 URL: https://issues.apache.org/jira/browse/BEAM-6243
>             Project: Beam
>          Issue Type: Bug
>          Components: runner-flink
>            Reporter: Robert Bradshaw
>            Priority: Major
>          Time Spent: 40m
>  Remaining Estimate: 0h
>
> The elements in TFX intermediate collections are dictionaries of (typically 
> single-element) numpy arrays, which are (relatively) expensive to serialize 
> (e.g. using pickle for the numpy wrapper of a primitive int/float, repeating 
> the column names in every element).
> Though it'd be good to use a better intermediate representation, this is 
> exacerbated because the fusion algorithm does not pack as much possible into 
> executable stages. 



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to