You are viewing a plain text version of this content. The canonical link for it is here.

Posted to user@storm.apache.org by Dominik Safaric <do...@gmail.com> on 2016/11/16 20:50:59 UTC

Having the Nimbus and a worker on the same node of the cluster

Hi,

It is generally recommended for streaming engines, also including Storm to run a separate master node - in the case of Storm, the Nimbus. 

However, why should one in Storm run the Nimbus on a separate node? 

Performance wise, the Nimbus isn’t intense unlike of course workers. 

In terms of fault tolerance and a failing Nimbus, semantically there is no difference. 

Hence, what are the main reasons behind this rationale? 

Thanks in advance.