Bitmap index scan behavior in PostgreSQL - Step-by-Step Execution

Concept Flow - Bitmap index scan behavior

Start Query

↓

Use Index to Find Matching Rows

↓

Build Bitmap of Row Locations

↓

Combine Bitmaps if Multiple Indexes

↓

Scan Heap Using Bitmap

↓

Return Matching Rows

↓

End

The query planner uses indexes to find matching rows, builds a bitmap of their locations, then scans the table efficiently using this bitmap to return results.

Execution Sample

PostgreSQL

EXPLAIN (ANALYZE, BUFFERS) SELECT * FROM employees WHERE department_id = 5;

This query uses a bitmap index scan to find all employees in department 5 efficiently.

Execution Table

Step	Action	Details	Result
1	Start Query	PostgreSQL receives SELECT with WHERE department_id=5	Query begins
2	Index Scan	Use index on department_id to find matching row pointers	Found row pointers for matching rows
3	Build Bitmap	Create bitmap of matching row locations	Bitmap created with row locations
4	Heap Scan	Scan table rows using bitmap to fetch actual data	Rows fetched from heap
5	Return Rows	Return matching rows to client	Result set sent
6	End	Query execution complete	All matching rows returned

💡 All matching rows found and returned, query completes

Variable Tracker

Variable	Start	After Step 2	After Step 3	After Step 4	Final
Bitmap	empty	row pointers collected	bitmap built	used to scan heap	cleared after scan
Result Set	empty	empty	empty	rows fetched	rows returned

Key Moments - 2 Insights

Why does PostgreSQL build a bitmap instead of fetching rows immediately after index scan?

Because the bitmap collects all matching row locations first, allowing a single efficient heap scan instead of many random accesses. See execution_table rows 3 and 4.

What happens if multiple indexes are used in the bitmap index scan?

PostgreSQL combines bitmaps from each index to find the intersection or union of matching rows before scanning the heap. This is part of the bitmap build step (row 3).

Visual Quiz - 3 Questions

Look at the execution_table, at which step is the bitmap actually used to fetch rows from the table?

A Step 4 - Heap Scan

B Step 3 - Build Bitmap

C Step 2 - Index Scan

D Step 5 - Return Rows

💡 Hint

Check the 'Action' and 'Details' columns in execution_table row 4.

According to variable_tracker, what is the state of the Result Set after Step 3?

A Contains all matching rows

B Still empty, rows not fetched yet

C Partially filled with some rows

D Cleared after heap scan

💡 Hint

Look at the 'Result Set' row under 'After Step 3' in variable_tracker.

If the bitmap was not built and rows were fetched immediately after index scan, what would likely happen?

A Query would be faster due to immediate fetch

B Heap scan would be more efficient

C Many random disk accesses would slow down query

D No change in performance

💡 Hint

Refer to key_moments explanation about why bitmap is built before heap scan.

Concept Snapshot

Bitmap Index Scan in PostgreSQL:
- Uses index to find matching row locations
- Builds a bitmap of these locations
- Scans table heap once using bitmap
- Efficient for many matching rows
- Combines multiple bitmaps if needed

Full Transcript

In PostgreSQL, a bitmap index scan helps find rows matching a condition efficiently. First, the index is scanned to find row locations matching the query. These locations are stored in a bitmap, which is a compact map of rows to fetch. Then, PostgreSQL scans the table heap once using this bitmap to retrieve the actual rows. This reduces random disk access and speeds up queries with many matches. If multiple indexes are involved, their bitmaps are combined before scanning the heap. The process ends by returning all matching rows to the client.

Practice

1. What is the main purpose of a Bitmap Index Scan in PostgreSQL?

easy

A. To delete rows using bitmap operations

B. To create a bitmap of matching row locations from indexes for efficient retrieval

C. To update rows in the table based on index values

D. To directly fetch rows from the table without using indexes

Solution

1. Step 1: Understand Bitmap Index Scan role

   Bitmap Index Scan creates a bitmap representing matching row positions using indexes.

2. Step 2: Differentiate from other scans

   It does not fetch rows directly but prepares a bitmap for efficient row retrieval later.

3. Final Answer:

   To create a bitmap of matching row locations from indexes for efficient retrieval -> Option B

4. Quick Check:

   Bitmap Index Scan = bitmap of row locations [OK]

Hint: Bitmap Index Scan builds a map of rows to fetch [OK]

Common Mistakes:

• Confusing bitmap scan with direct table scan
• Thinking bitmap scan updates or deletes rows
• Assuming bitmap scan fetches rows immediately

2. Which of the following is the correct syntax to perform a Bitmap Index Scan in a PostgreSQL EXPLAIN query output?

easy

A. Index Scan on table_name (cost=0.29..8.31 rows=5 width=12)

B. Bitmap Heap Scan on index_name (cost=0.29..8.31 rows=5 width=12)

C. Bitmap Index Scan on index_name (cost=0.29..8.31 rows=5 width=12)

D. Seq Scan on index_name (cost=0.29..8.31 rows=5 width=12)

Solution

1. Step 1: Identify Bitmap Index Scan syntax

   Bitmap Index Scan appears as "Bitmap Index Scan on index_name" in EXPLAIN output.

2. Step 2: Differentiate from other scans

   Bitmap Heap Scan fetches rows using bitmap, Index Scan and Seq Scan are different methods.

3. Final Answer:

   Bitmap Index Scan on index_name (cost=0.29..8.31 rows=5 width=12) -> Option C

4. Quick Check:

   Bitmap Index Scan syntax = Bitmap Index Scan on index_name (cost=0.29..8.31 rows=5 width=12) [OK]

Hint: Look for 'Bitmap Index Scan on' in EXPLAIN output [OK]

Common Mistakes:

• Confusing Bitmap Heap Scan with Bitmap Index Scan
• Choosing Index Scan or Seq Scan syntax incorrectly
• Misreading EXPLAIN output keywords

3. Given a table employees with an index on department_id, what will the Bitmap Index Scan do when you run:
EXPLAIN SELECT * FROM employees WHERE department_id = 5;?

medium

A. It creates a bitmap of row locations where department_id = 5, then fetches those rows efficiently

B. It scans the entire table sequentially without using the index

C. It updates the rows where department_id = 5

D. It deletes rows where department_id = 5

Solution

1. Step 1: Understand Bitmap Index Scan on condition

   The scan uses the index on department_id to find matching rows and creates a bitmap of their locations.

2. Step 2: Explain how rows are fetched

   Using the bitmap, it fetches only those rows efficiently from the table, avoiding full scan.

3. Final Answer:

   It creates a bitmap of row locations where department_id = 5, then fetches those rows efficiently -> Option A

4. Quick Check:

   Bitmap Index Scan finds matching rows then fetches [OK]

Hint: Bitmap Index Scan finds and fetches matching rows efficiently [OK]

Common Mistakes:

• Thinking it scans the whole table sequentially
• Confusing scan with update or delete operations
• Assuming it fetches rows without bitmap

4. You see a query plan with Bitmap Index Scan followed by Bitmap Heap Scan, but the query is running very slowly. What could be a likely cause?

medium

A. The index does not exist on the queried column

B. The table is empty, so no rows are fetched

C. The query is missing a WHERE clause

D. The bitmap is too large because the condition matches too many rows, causing inefficient heap fetch

Solution

1. Step 1: Analyze Bitmap Index Scan and Bitmap Heap Scan behavior

   Bitmap Index Scan creates a bitmap of matching rows; Bitmap Heap Scan fetches rows using that bitmap.

2. Step 2: Understand performance impact of large bitmap

   If too many rows match, the bitmap is large, causing many random disk accesses and slowing the query.

3. Final Answer:

   The bitmap is too large because the condition matches too many rows, causing inefficient heap fetch -> Option D

4. Quick Check:

   Large bitmap = slow Bitmap Heap Scan [OK]

Hint: Large bitmap means many rows matched, slowing fetch [OK]

Common Mistakes:

• Assuming missing index causes Bitmap Index Scan
• Thinking missing WHERE clause causes Bitmap Index Scan
• Believing empty table causes slow scan

5. You want to optimize a query that uses Bitmap Index Scan but runs slowly because it matches many rows. Which approach can improve performance?

hard

A. Add more selective WHERE conditions to reduce matching rows before Bitmap Index Scan

B. Drop the index to force a sequential scan

C. Increase the work_mem setting to allow larger bitmaps in memory

D. Rewrite the query to use a JOIN instead of WHERE clause

Solution

1. Step 1: Understand Bitmap Index Scan memory usage

   Bitmap Index Scan builds a bitmap in memory; if too large, it spills to disk, slowing performance.

2. Step 2: Improve performance by adding selective conditions

   Adding more selective WHERE conditions reduces matching rows, making bitmap smaller and faster.

3. Step 3: Evaluate other options

   Increasing work_mem helps but may not be sufficient; dropping index or rewriting query may not improve bitmap scan efficiency.

4. Final Answer:

   Add more selective WHERE conditions to reduce matching rows before Bitmap Index Scan -> Option A

5. Quick Check:

   More selective WHERE = smaller bitmap = faster scan [OK]

Hint: Add selective WHERE clauses to reduce bitmap size [OK]

Common Mistakes:

• Dropping index thinking it helps performance
• Assuming increasing work_mem always solves slowness
• Believing rewriting query always improves bitmap scan