Dynamic fields and slow queries - performance

Question

Currently, I'm managing a set of lists containing a number of members.
Every list can look different, when it comes to fields and the naming of these fields.
Typically, a basic list member could look like so (from my members collection):
{
"_id" : ObjectId("52284ae408edcb146200009f"),
"list_id" : 1,
"status" : "active",
"imported" : 1,
"fields" : {
"firstname" : "John",
"lastname" : "Doe",
"email" : "john#example.com",
"birthdate" : ISODate("1977-09-03T23:08:20.000Z"),
"favorite_color" : "Green",
"interests" : [
{
"id" : 8,
"value" : "Books"
},
{
"id" : 10,
"value" : "Travel"
},
{
"id" : 12,
"value" : "Cooking"
},
{
"id" : 15,
"value" : "Wellnes"
}
]
},
"created_at" : ISODate("2012-05-06T15:12:26.000Z"),
"updated_at" : ISODate("2012-05-06T15:12:26.000Z")
}
All the fields under the "fields" index, is fields that is unique for the current list id - and these fields can change for every list ID, which means a new list could look like so:
{
"_id" : ObjectId("52284ae408edcb146200009f"),
"list_id" : 2,
"status" : "active",
"imported" : 1,
"fields" : {
"fullname" : "John Doe",
"email" : "john#example.com",
"cell" : 123456787984
},
"created_at" : ISODate("2012-05-06T15:12:26.000Z"),
"updated_at" : ISODate("2012-05-06T15:12:26.000Z")
}
Currently, my application is allowing users to search dynamically in each of the customs fields, but since they have no indexes, this process can be very slow.
I don't believe it's an option to allow list creaters to select which fields should be indexed - but I really need to speed this up.
Is there any solution for this?

Answer 1

If you refactor your documents in a way that you have an array of fields, you can leverage indexes.
fields: [
{ name: 'fullName', value: 'John Doe' },
{ name: 'email', value: 'john#example.com' },
...
]
Create an index on fields.name and fields.value.
Of course this is not a solution for "deeper" values like your interests list.