How Elasticsearch Works – Understanding Indexing, Documents, and Search

In the previous lesson, you learned why Elasticsearch is needed and how it overcomes the limitations of traditional databases.

Now it's time to understand what happens behind the scenes.

Have you ever wondered:

• How does Elasticsearch find information so quickly?
• Where is the data stored?
• What exactly is an Index?
• What is a Document?
• How does a search happen in just a few milliseconds?

By the end of this lesson, you'll understand the core building blocks of Elasticsearch and how they work together.

Whenever new data is added to Elasticsearch, it follows a simple process:

Application

│

▼

Send Data

│

▼

Elasticsearch

│

▼

Store as a Document

│

▼

Create an Index

│

▼

Data Becomes Searchable

Instead of searching every document one by one, Elasticsearch creates a special searchable structure called an index, allowing it to locate information much faster.

A Document is the smallest unit of data stored in Elasticsearch.

Think of a document as one complete record.

Real-Life Example

Imagine an online shopping website.

One product might contain:

• Product Name
• Price
• Category
• Brand
• Rating

Each product is stored as one document.

For example:

{

"product_name": "Wireless Mouse",

"brand": "Logitech",

"price": 999,

"category": "Electronics",

"rating": 4.7

}

This entire JSON object is called a document.

Elasticsearch stores data in JSON (JavaScript Object Notation).

JSON is:

• Easy for humans to read.
• Easy for computers to process.
• Supported by most programming languages.

Almost every application that communicates with Elasticsearch sends and receives JSON data.

An Index is a collection of related documents.

Think of it as a folder that stores similar types of information.

Example

Suppose you're building an e-commerce application.

Instead of putting everything into one place, you organize the data like this:

Products Index

├── Laptop

├── Mouse

├── Keyboard

Customers Index

├── Customer 1

├── Customer 2

Orders Index

├── Order 101

├── Order 102

Each index contains documents related to a specific category.

Imagine a library.

The library itself contains many sections:

• Science
• History
• Fiction
• Technology

Each section is like an Index.

Inside every section are many books.

Each book is like a Document.

Library

│

├── Science (Index)

│      ├── Book 1 (Document)

│      ├── Book 2

│

├── Fiction (Index)

│      ├── Book 1

│      ├── Book 2

This organization makes it much easier to locate information.

Indexing is the process of preparing data so it can be searched efficiently.

When a new document is added, Elasticsearch:

Step 1

Receives the JSON document.

↓

Step 2

Analyzes the text.

↓

Step 3

Breaks the text into smaller searchable pieces called tokens.

↓

Step 4

Stores these tokens in its search index.

↓

Step 5

Makes the document searchable.

Suppose you add this product:

{

"product_name": "Wireless Bluetooth Headphones"

}

Elasticsearch analyzes the text and breaks it into words such as:

Wireless

Bluetooth

Headphones

These words are stored in the index.

Now users can search using words like:

• Wireless
• Bluetooth
• Headphones

and Elasticsearch can quickly find the matching document.

Suppose a customer searches for:

Bluetooth Headphones

Elasticsearch follows these steps:

User Searches

│

▼

Analyze Search Text

│

▼

Look Inside the Index

│

▼

Find Matching Documents

│

▼

Calculate Relevance Score

│

▼

Display Results

Notice that Elasticsearch searches the index, not every document individually.

This is one of the main reasons it is so fast.

Imagine you have 50,000 books.

Without an Index

You open every book until you find the word.

This could take hours.

With an Index

You first check the alphabetical index.

It immediately tells you which books contain the word.

You go directly to those books.

Elasticsearch works in a similar way.

When Elasticsearch analyzes text, it divides it into smaller pieces called tokens.

Example:

Learning Elasticsearch is Easy

After tokenization:

Learning

Elasticsearch

Easy

Common words like is, the, and a may be ignored, depending on the analyzer being used.

This helps improve search performance and relevance.

An Analyzer prepares text before it is indexed or searched.

It can:

• Convert text to lowercase.
• Remove punctuation.
• Split sentences into words.
• Remove common words.
• Apply language-specific processing.

For example:

Input:

Wireless Bluetooth Headphones

Output:

wireless

bluetooth

headphones

This allows searches to work even if the user types:

BLUETOOTH HEADPHONES

The case difference doesn't affect the search results.