[
https://issues.apache.org/jira/browse/BEAM-10475?focusedWorklogId=513673&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-513673
]
ASF GitHub Bot logged work on BEAM-10475:
-----------------------------------------
Author: ASF GitHub Bot
Created on: 18/Nov/20 18:03
Start Date: 18/Nov/20 18:03
Worklog Time Spent: 10m
Work Description: nehsyc commented on a change in pull request #13292:
URL: https://github.com/apache/beam/pull/13292#discussion_r526309131
##########
File path: sdks/python/apache_beam/transforms/util.py
##########
@@ -780,6 +783,48 @@ def expand(self, pcoll):
self.max_buffering_duration_secs,
self.clock))
+ @typehints.with_input_types(Tuple[K, V])
+ @typehints.with_output_types(Tuple[K, Iterable[V]])
+ class WithShardedKey(PTransform):
+ """A GroupIntoBatches transform that outputs batched elements associated
+ with sharded input keys.
+
+ The sharding is determined by the runner to balance the load during the
Review comment:
In Dataflow, the load balancing is done inside Windmill during the
shuffle before the `GroupIntoBatchesDoFn`. Basically Windmill can re-distribute
the input element to different workers and assign them different shard ids
accordingly.
Different runner may choose different strategies by overriding the default
implementation here. We will add the transform to the runner api (in follow-up
PRs) so runners are able to recognize the transform and do something special if
they want.
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 513673)
Time Spent: 18.5h (was: 18h 20m)
> GroupIntoBatches with Runner-determined Sharding
> ------------------------------------------------
>
> Key: BEAM-10475
> URL: https://issues.apache.org/jira/browse/BEAM-10475
> Project: Beam
> Issue Type: Improvement
> Components: runner-dataflow
> Reporter: Siyuan Chen
> Assignee: Siyuan Chen
> Priority: P2
> Labels: GCP, performance
> Time Spent: 18.5h
> Remaining Estimate: 0h
>
> [https://s.apache.org/sharded-group-into-batches|https://s.apache.org/sharded-group-into-batches__]
> Improve the existing Beam transform, GroupIntoBatches, to allow runners to
> choose different sharding strategies depending on how the data needs to be
> grouped. The goal is to help with the situation where the elements to process
> need to be co-located to reduce the overhead that would otherwise be incurred
> per element, while not losing the ability to scale the parallelism. The
> essential idea is to build a stateful DoFn with shardable states.
>
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
