Sequential scan vs index scan in PostgreSQL - When to Use Which

The Big Idea

What if you could find any piece of data instantly without flipping through everything?

The Scenario

Imagine you have a huge phone book and you want to find all people named "John." You start at the first page and look at every single name until you find all the Johns.

The Problem

This method is very slow and tiring because you check every name, even if the Johns are only on a few pages. It wastes time and energy, especially when the phone book is very big.

The Solution

Using an index is like having a special list that tells you exactly which pages have the name "John." You can jump straight to those pages without flipping through the whole book.

Before vs After

✗ Before

SELECT * FROM people WHERE name = 'John'; -- scans all rows

✓ After

CREATE INDEX idx_name ON people(name);
SELECT * FROM people WHERE name = 'John'; -- uses index to find rows faster

What It Enables

This lets the database find data quickly and efficiently, saving time and resources.

Real Life Example

When you search for a product on an online store, the system uses indexes to quickly show matching items instead of checking every product one by one.

Key Takeaways

Sequential scan checks every row one by one, which is slow for big data.

Index scan uses a special shortcut to find data faster.

Indexes make searching large databases much quicker and efficient.

Practice

1. What does a sequential scan do in PostgreSQL?

easy

A. Reads only the first 10 rows of a table

B. Uses an index to find specific rows quickly

C. Deletes rows based on a condition

D. Reads every row in the table one by one

Solution

1. Step 1: Understand sequential scan behavior

   A sequential scan reads all rows in the table from start to end without using any index.

2. Step 2: Compare with other options

   Using an index is an index scan, reading only first 10 rows or deleting rows are unrelated actions.

3. Final Answer:

   Reads every row in the table one by one -> Option D

4. Quick Check:

   Sequential scan = full table read [OK]

Hint: Sequential scan reads all rows, no index used [OK]

Common Mistakes:

• Confusing sequential scan with index scan
• Thinking sequential scan reads only some rows
• Assuming sequential scan deletes rows

2. Which of the following is the correct way to see if PostgreSQL uses an index scan or sequential scan?

easy

A. SELECT * FROM table WHERE id = 1;

B. EXPLAIN SELECT * FROM table WHERE id = 1;

C. CREATE INDEX ON table(id);

D. DROP INDEX index_name;

Solution

1. Step 1: Identify command to check query plan

   The EXPLAIN command shows how PostgreSQL executes a query, including scan type.

2. Step 2: Eliminate other options

   SELECT runs query but doesn't show plan; CREATE and DROP INDEX modify indexes, not show plans.

3. Final Answer:

   EXPLAIN SELECT * FROM table WHERE id = 1; -> Option B

4. Quick Check:

   EXPLAIN shows scan type [OK]

Hint: Use EXPLAIN before query to see scan type [OK]

Common Mistakes:

• Running SELECT without EXPLAIN to check scan
• Confusing index creation with scan checking
• Using DROP INDEX to check scans

3. Given a table users(id SERIAL PRIMARY KEY, name TEXT) with 1 million rows, which scan is PostgreSQL likely to use for this query?
SELECT * FROM users WHERE id = 500000;

medium

A. Index scan using the primary key index

B. Sequential scan scanning all 1 million rows

C. Bitmap heap scan reading random rows

D. No scan, query will fail

Solution

1. Step 1: Analyze query condition and table size

   The query filters by primary key id, which has an index, and the table is large (1 million rows).

2. Step 2: Determine efficient scan type

   PostgreSQL uses an index scan to quickly find the single matching row instead of scanning all rows.

3. Final Answer:

   Index scan using the primary key index -> Option A

4. Quick Check:

   Selective query on indexed column = index scan [OK]

Hint: Selective query on indexed column uses index scan [OK]

Common Mistakes:

• Assuming sequential scan for large tables with indexed filter
• Confusing bitmap heap scan with index scan
• Thinking query fails without reason

4. You wrote this query:
SELECT * FROM orders WHERE customer_id = 123;
But EXPLAIN shows a sequential scan instead of an index scan. What could be the reason?

medium

A. There is no index on customer_id

B. The table is empty

C. The query syntax is incorrect

D. PostgreSQL always uses sequential scan

Solution

1. Step 1: Check index presence on filter column

   If no index exists on customer_id, PostgreSQL must scan all rows sequentially.

2. Step 2: Evaluate other options

   Empty table would still show scan but no rows; syntax is correct; PostgreSQL chooses scan type based on indexes.

3. Final Answer:

   There is no index on customer_id -> Option A

4. Quick Check:

   No index on filter column = sequential scan [OK]

Hint: No index on filter column causes sequential scan [OK]

Common Mistakes:

• Assuming syntax error causes scan type
• Thinking PostgreSQL always uses sequential scan
• Ignoring missing index as cause

5. You have a large table products with millions of rows and an index on category_id. You run:
SELECT * FROM products WHERE category_id IN (1, 2, 3, 4, 5);
PostgreSQL chooses a sequential scan instead of an index scan. Why might this happen?

hard

A. The index on category_id is corrupted

B. Sequential scan is always faster for any query

C. The query is not selective enough; many rows match

D. The IN clause is invalid syntax

Solution

1. Step 1: Understand query selectivity

   The query matches many rows because it filters on multiple categories, reducing selectivity.

2. Step 2: Explain PostgreSQL scan choice

   When many rows match, PostgreSQL prefers sequential scan as it can be faster than many index lookups.

3. Final Answer:

   The query is not selective enough; many rows match -> Option C

4. Quick Check:

   Low selectivity = sequential scan preferred [OK]

Hint: Low selectivity queries often use sequential scan [OK]

Common Mistakes:

• Assuming index corruption without evidence
• Thinking sequential scan is always slower
• Believing IN clause is invalid syntax