This might be a good question for the ES-forum, since they do more data modeling.
But here's how I start:
- every row is a document (ie. JSON object)
- each column is a field (ie. JSON property)
To handle both sets, you could use two approaches:
- store each collection of documents in a different index (so an index for sheet-old and sheet-new).
- store in the same index, but add json-property to each document indicating to which excel-sheet it belongs.