Embedding vs referencing decision in MongoDB - Performance Comparison

Time Complexity: Embedding vs referencing decision

O(n)

Understanding Time Complexity

When choosing between embedding and referencing in MongoDB, it's important to understand how the time to get data changes as your data grows.

We want to know how the way data is stored affects how long queries take when the amount of data increases.

Scenario Under Consideration

Analyze the time complexity of fetching related data using embedding vs referencing.

// Embedding example
const user = db.users.findOne({ _id: userId });
// user document contains embedded posts array

// Referencing example
const user = db.users.findOne({ _id: userId });
const posts = db.posts.find({ userId: user._id }).toArray();
    

This code shows two ways to get a user's posts: either embedded inside the user document or stored separately and linked by userId.

Identify Repeating Operations

Look at what repeats when fetching posts for a user.

• Primary operation: Reading posts data either from embedded array or separate collection.
• How many times: For embedding, posts are read once inside user document. For referencing, a separate query fetches all posts for the user.

How Execution Grows With Input

Consider how the number of posts (n) affects query time.

Input Size (n)	Approx. Operations
10 posts	Embedding: 1 read including 10 posts; Referencing: 1 user read + 1 query returning 10 posts
100 posts	Embedding: 1 read including 100 posts; Referencing: 1 user read + 1 query returning 100 posts
1000 posts	Embedding: 1 read including 1000 posts; Referencing: 1 user read + 1 query returning 1000 posts

Pattern observation: Both methods read all posts, but embedding reads them all at once inside one document, while referencing requires a separate query that grows with the number of posts.

Final Time Complexity

Time Complexity: O(n)

This means the time to fetch posts grows linearly with the number of posts, whether embedded or referenced.

Common Mistake

[X] Wrong: "Embedding always makes queries faster because all data is in one place."

[OK] Correct: If the embedded data grows very large, reading the whole document can be slow and use more memory. Referencing can be better for very large or frequently changing related data.

Interview Connect

Understanding how embedding and referencing affect query time helps you explain design choices clearly and shows you think about how data size impacts performance.

Self-Check

"What if we added an index on the referencing field? How would that change the time complexity of fetching posts?"

Practice

1. Which scenario is best suited for embedding related data in MongoDB?

easy

A. Storing a user's profile and their list of addresses when addresses are few and rarely shared.

B. Storing product details and linking to multiple categories by ID.

C. Storing large documents that are frequently updated independently.

D. Storing shared tags used by many blog posts.

Solution

1. Step 1: Understand embedding use case

   Embedding is best when related data is small and tightly coupled, like a user's addresses.

2. Step 2: Analyze options for embedding

   Storing a user's profile and their list of addresses when addresses are few and rarely shared fits because addresses are few and not shared, so embedding improves read speed.

3. Final Answer:

   Storing a user's profile and their list of addresses when addresses are few and rarely shared. -> Option A

4. Quick Check:

   Embedding = small, tightly related data [OK]

Hint: Embed when data is small and tightly linked [OK]

Common Mistakes:

• Embedding large or shared data causing duplication
• Referencing small, simple data unnecessarily

2. Which of the following is the correct way to reference a document in MongoDB?

easy

A. { "author_id": ObjectId("507f1f77bcf86cd799439011") }

B. { "author": { "$ref": "users", "$id": ObjectId("507f1f77bcf86cd799439011") } }

C. { "author": ObjectId("507f1f77bcf86cd799439011") }

D. { "author": "507f1f77bcf86cd799439011" }

Solution

1. Step 1: Identify referencing syntax

   Referencing stores the related document's ID as a field, usually named with _id suffix.

2. Step 2: Check options for correct ID reference

   { "author_id": ObjectId("507f1f77bcf86cd799439011") } correctly stores the ObjectId with a clear field name author_id.

3. Final Answer:

   { "author_id": ObjectId("507f1f77bcf86cd799439011") } -> Option A

4. Quick Check:

   Reference = store related document's ID [OK]

Hint: Reference by storing ObjectId in a field ending with _id [OK]

Common Mistakes:

• Using string instead of ObjectId for references
• Using $ref and $id without DBRef context
• Embedding full document instead of referencing

3. Given a MongoDB collection orders where each order embeds a small array of items, what is the expected result of querying an order by ID?

medium

A. Only the order document without any items data.

B. An error because embedded arrays cannot be queried.

C. The order document with all embedded items included in the result.

D. A list of item documents from a separate collection.

Solution

1. Step 1: Understand embedding query behavior

   When data is embedded, querying the parent document returns all embedded data inside it.

2. Step 2: Analyze query result for embedded items

   Querying by order ID returns the order document including the embedded items array.

3. Final Answer:

   The order document with all embedded items included in the result. -> Option C

4. Quick Check:

   Embedding returns full embedded data [OK]

Hint: Embedded data returns inside parent document query [OK]

Common Mistakes:

• Expecting separate documents for embedded data
• Thinking embedded arrays cause query errors
• Confusing referencing with embedding results

4. You have a MongoDB schema where user documents embed an array of posts. You notice performance issues when updating posts frequently. What is the likely cause and fix?

medium

A. Indexing embedded arrays causes slow updates; remove indexes.

B. Referencing posts causes slow updates; embed posts instead.

C. Using ObjectId for posts causes slow updates; use strings instead.

D. Embedding large arrays causes slow updates; switch to referencing posts separately.

Solution

1. Step 1: Identify embedding impact on updates

   Embedding large or frequently updated arrays causes the whole document to be rewritten, slowing updates.

2. Step 2: Choose referencing to improve update performance

   Referencing posts separately allows updating posts without rewriting the entire user document.

3. Final Answer:

   Embedding large arrays causes slow updates; switch to referencing posts separately. -> Option D

4. Quick Check:

   Large/frequent updates = use referencing [OK]

Hint: Large, frequently updated data? Use referencing [OK]

Common Mistakes:

• Embedding large arrays despite frequent updates
• Ignoring update performance impact
• Misunderstanding ObjectId role in updates

5. You are designing a MongoDB schema for a blogging platform. Each blog post has many comments, and comments can be edited or deleted independently. Also, comments can be very large in number. What is the best design choice?

hard

A. Embed all comments inside each blog post document.

B. Reference comments in a separate collection linked by post ID.

C. Embed only the first 10 comments and reference the rest.

D. Store comments as a JSON string inside the blog post.

Solution

1. Step 1: Analyze comment characteristics

   Comments are numerous, large, and updated independently, so embedding would cause large documents and slow updates.

2. Step 2: Choose referencing for scalability and independent updates

   Referencing comments in a separate collection allows efficient updates and avoids large blog post documents.

3. Final Answer:

   Reference comments in a separate collection linked by post ID. -> Option B

4. Quick Check:

   Large, independent data = referencing [OK]

Hint: Large, independent data? Reference separately [OK]

Common Mistakes:

• Embedding large, frequently updated arrays
• Mixing embedding and referencing without clear rules
• Storing JSON strings instead of structured data