[
https://issues.apache.org/jira/browse/FLINK-1725?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14727399#comment-14727399
]
ASF GitHub Bot commented on FLINK-1725:
---------------------------------------
Github user gdfm commented on the pull request:
https://github.com/apache/flink/pull/1069#issuecomment-137096426
@tillrohrmann the reason why 2 is a "magic number" is the following.
Immagine channels as containers, and the load (the incoming data stream) as
a liquid.
When a key is split in two channels, it creates a "bridge" between these
channels.
You can imagine it as a pipe between the two specific containers that
enables sharing the load, and brings the liquid to the same level (this happens
thanks to the fact that the new messages are sent to the least loaded of the
two containers).
Now, if you have enough of these pipes between pairs of containers, you
will effectively establish a network of load sharing among them. Any increase
in pressure on one of the containers can be shared across the network
effectively, which brings the load balance.
The trick is to have "enough" keys to create enough pipes.
> New Partitioner for better load balancing for skewed data
> ---------------------------------------------------------
>
> Key: FLINK-1725
> URL: https://issues.apache.org/jira/browse/FLINK-1725
> Project: Flink
> Issue Type: Improvement
> Components: New Components
> Affects Versions: 0.8.1
> Reporter: Anis Nasir
> Assignee: Anis Nasir
> Labels: LoadBalancing, Partitioner
> Original Estimate: 336h
> Remaining Estimate: 336h
>
> Hi,
> We have recently studied the problem of load balancing in Storm [1].
> In particular, we focused on key distribution of the stream for skewed data.
> We developed a new stream partitioning scheme (which we call Partial Key
> Grouping). It achieves better load balancing than key grouping while being
> more scalable than shuffle grouping in terms of memory.
> In the paper we show a number of mining algorithms that are easy to implement
> with partial key grouping, and whose performance can benefit from it. We
> think that it might also be useful for a larger class of algorithms.
> Partial key grouping is very easy to implement: it requires just a few lines
> of code in Java when implemented as a custom grouping in Storm [2].
> For all these reasons, we believe it will be a nice addition to the standard
> Partitioners available in Flink. If the community thinks it's a good idea, we
> will be happy to offer support in the porting.
> References:
> [1].
> https://melmeric.files.wordpress.com/2014/11/the-power-of-both-choices-practical-load-balancing-for-distributed-stream-processing-engines.pdf
> [2]. https://github.com/gdfm/partial-key-grouping
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
