Store/Sync Facebook Graph in a NoSQL (MongoDB)

Question

I'm building an app that needs to be able to extend facebook graph data.

I'm new to NoSQL storage and i'm looking for help.

Using the graph api, i can retreive a user, since i would like my app to be able to extend several social graph providers, i move every specific facebook keys retreived into a facebook array subset.

[User] => Array
    (
        [_id] => 4dd50c139bcb233c0c000000
        [name] => Foo Bar
        [first_name] => Foo 
        [last_name] => Bar
        [username] => fbar
        [location] => Array
            (
                [id] => 110774245616525
                [name] => Paris, France
            )

        [gender] => male
        [email] => [email protected]
        [timezone] => 2
        [locale] => fr_FR
        [facebook] => Array
            (
                [id] => 12345678
                [link] => http://www.facebook.com/foobar
                [verified] => 1
                [updated_time] => 2011-05-16T17:30:23+0000
                [picture] => https://graph.facebook.com/12345678/picture
            )

        [created] => MongoDate Object
            (
                [sec] => 1305807891
                [usec] => 0
            )
    )

• Is this good practice ?

Then i grab his friends, and i want to be able to keep them in sync with my database. I don't know if i should register each friend as separate users and try to use references, or if i can just add a Friend subset.

• What would have the best performance and the easiest to keep in sync ?

     [User] => Array
     (         
          [Friend] => Array
          (
              [0] => Array
                  (
                      [id] => 12345678
                      [name] => Foo Bar
                  )
  
              [1] => Array
                  (
                      [id] => 12345678
                      [name] => Foo Bar
                  )
  
              [2] => Array
                  (
                      [id] => 12345678
                      [name] => Foo Bar
                  )
  

• Problem rise to another level with FriendLists, how should i store them ?? embed everything (and have a ton of duplicates in my User) or use reference ? How should i do that ?

I read : http://www.mongodb.org/display/DOCS/Trees+in+MongoDB which is quite helpfull... But i'm still unsure what i should do.

Thanks a lot.

Answer 1

I had a similar application and I stored friends as an other user. But be sure that you have a flag that indicates if that user is an application user or not, or the data would be somehow chaotic at some point.

for user:

{
  fbid: xxxx,
  name : "xxxxx",
  ......
  friends : [ xxxx, xxxx, xxxx ],
  is_app_user : true
}

for each friend (who are not application user) :

{
  fbid: xxxx,
  name : "xxxxx",
  is_app_user : false
}

and when they also login you can make is_app_user : true for them aswell.

PS: dont forget to put unique index on fbid

Answer 2

This is a question of "embed vs. reference".

Here's one good reply on SO.

Here are the official docs on schema design. They're a good place to start looking at these questions.