Simple Data model of repository

[Mick]

Hi,

What would be the best way to represent in IDL a repository(id,stores) 
with many stores(id,content) considering efficiency and speed. Eg:
Repo1{
    id
    store1{
        id
        content
    }
    store2{
        id
        content
    }
    store3{
        id
        content
    }
}
I could put it in
1)
record store{
    string id;
    array<content> storeContent;
}
record repo{
    string id;
    array<store> stores;
}
2)
record store{
    string id;
    array<content> storeContent;
}

array<store> repo();

3)
record store{
    string id;
    array<content> storeContent;
}

map<store> repo();

Or maby there is some other schema that would be optimal regarding such 
simple data model?

Best,
Mick