You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@metron.apache.org by Simon Elliston Ball <si...@simonellistonball.com> on 2017/09/03 20:00:38 UTC
Shuffle methods in indexing and enrichment topologies
Is there a reason we use shuffle instead of local or shuffle in the indexing topology?
It seems like LOCAL_OR_SHUFFLE would help us get better pipelines through Storm and reduce the storm to storm shuffle traffic. Did I miss something here about balancing, or is it worth changing?
Simon