Marvel is installed as an Elasticsearch plugin. The plugin must be installed on every node in the cluster. By default, the plugin will store data in the same Elasticsearch cluster that it is monitoring.
If you are monitoring a production cluster we recommend that you send this data to another cluster that will serve as a monitoring cluster.
This will help to prevent a problem in one cluster from impacting the other cluster. For example, when your production cluster is having trouble you would like to make sure you can view the Marvel data easily.