Firebase Realtime Database - Best practice for additional data nodes (lookups and initial load)

Question

I have a Firebase Realtime Database with the below structure.

I wish to fetch all "notes" a user has access to and, initially, only show titles for the notes.

notes: {
  "noteId-1345" : {
    "access" : {
      "author": "1234567890"
        "members": {
          "1234567890": 0 <--- Author
          "0987654321": 1 <--- Member
        }
      },
    "data" : {
      "title": "Hello",
      "content": "Konichiwa!",
      "comment": "123"
    }
  }
}

To be able to fetch all notes a user has access to I have expanded my data model by keeping an additional user_notes node in the root: Whenever I associate a user (update of members) with a note, I write that relation both in /notes/$noteid and in /user_notes/$uid.

user_notes: {
  "$uid": {
    "noteId-1345": true
  }
}

When fetching initial data I only need the "notes" the user has access to - including titles.

(I only fetch the entire "note" if a user wants to view a complete "note")

I begin by fetching the ids for notes the user has access to and then I have to do another lookup to fetch the titles.

let titles = []
database.ref(`user_notes/${uid}`)
.on('value', (snaps) => {
  snaps.forEach((snap) => {
    const noteId = snap.key
    database.ref(`notes/${noteId}/data/title`).on('value', (noteSnap) => {
      const title = noteSnap.val()
      titles.push(title)
    }
  })
})

Is this the most efficient approach? - It seems inefficient to do double lookups.

Should I store title, and other data needed for initial load, in the user_notes node as well to avoid double lookups?

What is considered to be best practice in cases like this when using a NoSQL database?

Kind regards /K

Answer 1

What you're doing is indeed the common approach. It is not nearly as slow as you may initially think, since Firebase pipelines the requests over a single connection.

A few things to consider:

• I'd typically move the members for each note under a top-level node note_members. Separating the types of data typically makes it much easier to keep your security rules reasonable, and is documented under keep your data structure flat.

• If you'd like to get rid of the lookup, you can consider storing the title of each note under each user_notes node where you have the ID. You'd essentially replace the true with the name:

  user_notes: {
    "$uid": {
      "noteId-1345": "Hello"
    }
  }
  

  This simplifies the lookup code (its main advantage) and makes it a bit faster.

  This sort of data duplication is quite common when using Firebase and other NoSQL databases, you trade write complexity and extra data storage, for reach simplicity and scalability.