Elasticsearch document structure for performance optimization

Question

In Elasticsearch is it better to query on 50 fields in 1,000,000 documents or 1 field in 50,000,000 documents. I'll explain my problem with an example:

For case 1, our doc looks like:

doc1: {
    "f1": "abc",
    "f2": "pqr",
    "f3": "lmn",
    ...
    "f50": "xyz"
}

Here a should query will be performed from "f1" to "f50".

For case 2, our doc will look like:

doc1.1: {
    "f": "abc"
}

doc1.2: {
    "f": "pqr"
}    

doc1.3: {
    "f": "lmn"
}
...
...
doc1.50: {
    "f": "xyz"
}

Let's just say I am concerned about the docs containing the queried value irrespective of the field.

Answer 1

I would take the 50 million documents approach. ElasticSearch can handle that size easily with proper hardware (self experience), and it will be much easier for querying and data manipulation.