Advanced Patterns

Data Modeling - Advanced Patterns in NoSQL

The Toy Factory Story 🏭

Imagine you run a magical toy factory. Every day, thousands of orders come in. Kids want toys. Parents want them fast. You need to organize everything so toys get made and shipped without chaos.

NoSQL advanced patterns are like the clever tricks that make your factory super efficient.

Let’s learn four powerful tricks!

1. Aggregation Pattern 📊

What Is It?

Think of this: Every day, 1000 kids order toys. Instead of counting each order one by one when someone asks “How many toys did we sell today?”, you keep a running total.

The Aggregation Pattern pre-calculates summary data so you don’t have to count every time.

Simple Example

Without Aggregation (Slow): Every time the boss asks for today’s sales, a worker runs through ALL 1000 order papers.

With Aggregation (Fast): A special board at the entrance shows: “Today’s Sales: 1000” — updated every time an order comes in.

How It Works

// Order document
{
  "order_id": "A123",
  "product": "Teddy Bear",
  "price": 25,
  "date": "2024-01-15"
}

// Aggregation document (updated live)
{
  "date": "2024-01-15",
  "total_orders": 1000,
  "total_revenue": 25000,
  "top_product": "Teddy Bear"
}

When To Use It

• Dashboards showing totals
• Real-time counters
• Reports that need instant answers

The Magic Rule

Calculate once, read many times.

graph TD
    A["New Order Arrives"] --> B["Save Order"]
    B --> C["Update Summary Doc"]
    C --> D["Dashboard Shows Instantly"]

2. Bucket Pattern 🪣

What Is It?

Imagine your factory tracks every second of every machine. That’s millions of tiny notes!

The Bucket Pattern groups related small pieces into one container.

It’s like putting M&Ms into bags instead of carrying each one separately.

Simple Example

Without Bucket (Messy):

Document 1: Machine A, 10:00:01, Temperature: 75°
Document 2: Machine A, 10:00:02, Temperature: 76°
Document 3: Machine A, 10:00:03, Temperature: 75°
... 86,400 documents per day per machine!

With Bucket (Tidy):

{
  "machine": "Machine A",
  "date": "2024-01-15",
  "hour": 10,
  "readings": [
    {"second": 1, "temp": 75},
    {"second": 2, "temp": 76},
    {"second": 3, "temp": 75}
  ]
}

One document holds an hour of readings!

When To Use It

• Time-series data (sensor readings)
• Event logs
• Chat messages
• Activity streams

The Magic Rule

Group similar small things into logical containers.

graph TD
    A["Tiny Data Points"] --> B["Group by Time/Category"]
    B --> C["One Bucket Document"]
    C --> D["Fewer Documents"]
    D --> E["Faster Queries"]

3. Materialized Views 👁️

What Is It?

Your factory makes toys from many parts. A bike needs wheels, a frame, and handlebars. When customers ask “What bikes are available?”, you don’t want to check three different shelves.

A Materialized View is a pre-built answer sitting ready for questions.

It’s like having a “Ready Bikes” poster that shows complete bikes already assembled.

Simple Example

Original Collections:

// Products
{"_id": "bike1", "name": "Speed Bike"}

// Inventory
{"product_id": "bike1", "quantity": 50}

// Reviews
{"product_id": "bike1", "rating": 4.5}

Materialized View:

{
  "_id": "bike1",
  "name": "Speed Bike",
  "quantity": 50,
  "rating": 4.5,
  "in_stock": true
}

All the info combined! No need to look at three places.

How To Keep It Fresh

Two ways:

1. Update on change — When inventory changes, update the view
2. Update on schedule — Rebuild view every hour

When To Use It

• Product listings
• User profiles with stats
• Dashboard widgets
• Search results

The Magic Rule

Build the answer before anyone asks.

graph TD
    A["Products Collection"] --> D["Materialized View"]
    B["Inventory Collection"] --> D
    C["Reviews Collection"] --> D
    D --> E["Fast API Response"]

4. Data Duplication Strategy 📋

What Is It?

In a regular factory, you keep ONE master list of suppliers. But what if workers need supplier info while making toys? Running to the office takes time!

Data Duplication means copying some data where it’s needed most.

It’s like putting a copy of the supplier’s phone number right on each workstation.

Simple Example

Without Duplication:

// Order document
{
  "order_id": "A123",
  "customer_id": "C456"
}

// Customer collection (separate)
{
  "_id": "C456",
  "name": "Alice",
  "email": "alice@mail.com"
}

Need two lookups to show order with customer name!

With Duplication:

// Order document (includes customer name)
{
  "order_id": "A123",
  "customer_id": "C456",
  "customer_name": "Alice"
}

One lookup shows everything!

The Trade-off

When To Use It

• Data that rarely changes (names, categories)
• Data needed together often
• Read-heavy applications

When NOT To Duplicate

• Frequently changing data (prices, stock levels)
• Large data (images, files)
• Sensitive data (passwords)

The Magic Rule

Copy what you read together. Keep sources for updates.

graph TD
    A["Master Data"] --> B["Copy to Document A"]
    A --> C["Copy to Document B"]
    A --> D["Copy to Document C"]
    E["Update Master"] --> A
    A --> F["Sync Copies"]

Putting It All Together 🎯

Let’s see how our toy factory uses ALL four patterns:

Scenario: Order Processing System

// One powerful order document using ALL patterns

{
  // Basic order
  "order_id": "ORD-2024-001",
  "date": "2024-01-15",

  // DATA DUPLICATION: Customer info copied here
  "customer": {
    "id": "C456",
    "name": "Alice",
    "city": "New York"
  },

  // BUCKET PATTERN: Items grouped in arrays
  "items": [
    {"sku": "TOY-001", "name": "Teddy", "qty": 2},
    {"sku": "TOY-002", "name": "Robot", "qty": 1}
  ],

  // AGGREGATION: Pre-calculated totals
  "totals": {
    "item_count": 3,
    "subtotal": 75.00,
    "tax": 6.50,
    "total": 81.50
  }
}

// MATERIALIZED VIEW: Daily summary (separate doc)
{
  "view_type": "daily_summary",
  "date": "2024-01-15",
  "orders_count": 150,
  "revenue": 12225.00,
  "top_product": "Teddy Bear"
}

Quick Comparison Chart

Remember This! 🧠

1. Aggregation = Calculate ahead of time
2. Bucket = Group small things together
3. Materialized Views = Pre-build your answers
4. Duplication = Copy data where needed

All four patterns share one goal:

Make reading fast by doing work ahead of time.

You’ve Got This! 🚀

NoSQL patterns might seem tricky at first. But they’re just clever organizational tricks — like keeping your room tidy so you can find things fast.

Start with one pattern. Try it. See the speed improvement. Then add another.

Your databases will thank you. Your users will thank you. You’ll thank yourself!