Better ways to model this simple schema design (MongoDB)?

Question

I have the following case: A user can add contacts (who are other users) on the website. Optionally, a user may also organize his contacts in groups. A user can have many emails, addresses and phone numbers.

I thought of the following schema design (document store/mongodb). Are there any ways to improve this? My main concern is that the profile picture is embedded inside the document. I know that this is not a good practice, but for this particular purpose (assignment), I have to embed the picture (blob/gridfs) here as well. But I'm wondering how this schema can be improved.

How are you going to use the document? What sort of queries will you run? What sort of updates? — wdberkeley, Mar 18 '15 at 19:42
Thanks for your comment! I want to query the contacts of a particular user (get all the contacts from a user) and also show these contacts by groups (e.g. grab all the contacts that are in the group family). A user can update his own information and create his own groups (to later add contacts inside of it). — Moody, Mar 19 '15 at 08:47

score 0 · Accepted Answer · answered Mar 19 '15 at 15:14

For users, I think your current schema is fine. Keeping multiple addresses, phone numbers, and email addresses in arrays is good because there shouldn't be too many of them for a particular person and it's easy to query for "who has this email address" or "give me all the phone numbers of this person". You do seem to have a redundant e-mail field though - is that a special email, like the address for the account, that's distinguished from a contact email? If so I'd give it a descriptive name for the sake of other maintainers, a name like account_email. I wouldn't keep the photo as a blob, but you said that is some other requirement so I won't critique it.

I like the idea of doing contacts with groups using a separate contacts collection. I'd have a contacts collection with each document representing one contact

{
    "_id" : ObjectId("..."),
    "owner_id" : ObjectId("..."), // reference to user document of contact owner
    "contact_id" : ObjectId("..."), // reference to user document of contact
    "group" : "Rivals" // group name
}

Index on { "owner_id" : 1, "contact_id" : 1 } and maybe { "owner_id" : 1, "group" : 1 } and then queries like the following will be quick:

// get all contacts for user x
db.contacts.find({ "owner_id" : x })
// is user y a contact of user x?
db.contacts.count({ "owner_id" : x, "contact_id" : y }) != 0
// get all contacts in group "family" for user x
db.contacts.find({ "owner_id" : x, "group" : "family" })

Once you retrieve contacts, in order to retrieve human-friendly display information you will need to do a second query (application-level join) to retrieve the actual user documents of the contacts. If you want, you can denormalize some of the contact info into the contact document

{
    "_id" : ObjectId("..."),
    "owner_id" : ObjectId("..."), // reference to user document of contact owner
    "contact_id" : ObjectId("..."), // reference to user document of contact
    "group" : "Rivals", // group name
    "contact_name" : "Franke Frankers"
}

You won't need to do a second query if you include commonly needed info, but if the contact updates his/her information you may need to update every contact document referencing them.

I am using GridFS for the photo now by the way (no blob). If you think it can be done it a better way please let me know! :) — Moody, Mar 19 '15 at 22:55

Better ways to model this simple schema design (MongoDB)?

1 Answers1