I am familiar with the Mapping Explosion problem, but am unclear if it applies to fields that are not indexed in Lucene.
My intention is to store a properties object that has a high cardinality of keys (on the order of tens to hundreds of thousands per month). These fields are purely to be used for scripted aggregations and have no need to be searched against. I am unclear if we would run into the same issues performance issues as in the mapping explosion case, or if there is a storage method that makes them purely available for aggregation without the performance impact.