Performance Data for secondary Index

I was wondering how scalable the secondary Index is. I have a json doc which has
has as part of its structure a array of tag elements which store userid and name
of the user who was tagged like

Taggs
userid: 1, name: "Frank",
userid: 2, name: "Steve"
taggs

ignore the format its just to give id, so here is my question's

a) how do i access these tagg data ? doc.taggs.userid ?
b) lets assume i build a view what is the expected performance if i have like 10 million docs and
each doc can have between 1 and 1000 tagg's ? And how far will it scale ?
c )my query needs to find all taggs for a given user, would it be better to store all taggs for one user
in a separate doc and keep the docs in sync or is there a way to avoid this and still get a good response
time ?

Any ideas and pointers would be great

Thanks