On Fri, May 18, 2018 at 11:46 AM Raghu Angadi
&lt;rang...@google.com&gt; wrote:<br>&gt;<br>&gt; Thanks
Kenn.<br>&gt;<br>&gt; On Fri, May 18, 2018 at 11:02 AM Kenneth Knowles
&lt;k...@google.com&gt; wrote:<br>&gt;&gt;<br>&gt;&gt; The fact that
its usage has grown probably indicates that we have a large number of
transforms that can easily cause data loss /
duplication.<br>&gt;<br>&gt; Is this specific to Reshuffle or it is
true for any GroupByKey? I see Reshuffle as just a wrapper around
GBK.<br><br>The issue is when it&#39;s used in such a way that data
corruption can occur when the underlying GBK output is not stable.

Reply via email to