Creating partitioned tables in PostgreSQL - Performance & Efficiency

Time Complexity: Creating partitioned tables

O(n)

Understanding Time Complexity

When we create partitioned tables, we want to understand how the work grows as the data grows.

How does splitting data into parts affect the time to insert or query data?

Scenario Under Consideration

Analyze the time complexity of creating a partitioned table and inserting data.

CREATE TABLE sales (
  id SERIAL PRIMARY KEY,
  sale_date DATE NOT NULL,
  amount NUMERIC
) PARTITION BY RANGE (sale_date);

CREATE TABLE sales_2023 PARTITION OF sales
  FOR VALUES FROM ('2023-01-01') TO ('2024-01-01');

INSERT INTO sales (sale_date, amount) VALUES ('2023-06-15', 100);
    

This code creates a main table partitioned by date range and inserts a row into the correct partition.

Identify Repeating Operations

Look at what repeats when inserting many rows.

• Primary operation: Checking which partition a row belongs to.
• How many times: Once per inserted row.

How Execution Grows With Input

As you insert more rows, the system checks partitions for each row.

Input Size (n)	Approx. Operations
10	10 partition checks
100	100 partition checks
1000	1000 partition checks

Pattern observation: The number of partition checks grows directly with the number of rows inserted.

Final Time Complexity

Time Complexity: O(n)

This means the time to insert rows grows in a straight line with the number of rows.

Common Mistake

[X] Wrong: "Partitioning makes inserts instant no matter how many rows."

[OK] Correct: Each row still needs to be checked to find its partition, so time grows with rows.

Interview Connect

Understanding how partitioning affects time helps you design databases that handle large data smoothly.

Self-Check

"What if we added more partitions? How would that affect the time to find the right partition for each row?"

Practice

1. What is the main purpose of creating partitioned tables in PostgreSQL?

easy

A. To split a large table into smaller, manageable parts based on a column

B. To create multiple copies of the same table for backup

C. To combine several tables into one large table

D. To encrypt the data in a table for security

Solution

1. Step 1: Understand partitioning concept

   Partitioned tables divide a big table into smaller parts based on a column value, improving management and query performance.

2. Step 2: Compare options

   Backup, merging several tables, and encryption are not related to partitioning.

3. Final Answer:

   To split a large table into smaller, manageable parts based on a column -> Option A

4. Quick Check:

   Partitioned tables split big tables = A [OK]

Hint: Partitioning means splitting big tables by column values [OK]

Common Mistakes:

• Confusing partitioning with backup or encryption
• Thinking partitioning merges tables instead of splitting
• Assuming partitioning duplicates data

2. Which of the following is the correct syntax to create a partitioned table by range on column created_date?

easy

A. CREATE TABLE orders PARTITION BY LIST (created_date);

B. CREATE TABLE orders PARTITION ON RANGE (created_date);

C. CREATE TABLE orders PARTITION BY HASH (created_date);

D. CREATE TABLE orders PARTITION BY RANGE (created_date);

Solution

1. Step 1: Recall partition syntax

   PostgreSQL uses PARTITION BY RANGE (column) to create range partitions.

2. Step 2: Check options

   CREATE TABLE orders PARTITION BY RANGE (created_date); uses correct syntax. CREATE TABLE orders PARTITION ON RANGE (created_date); uses wrong keyword 'PARTITION ON'. PARTITION BY LIST and PARTITION BY HASH use different partition types.

3. Final Answer:

   CREATE TABLE orders PARTITION BY RANGE (created_date); -> Option D

4. Quick Check:

   Correct syntax uses PARTITION BY RANGE [OK]

Hint: Use PARTITION BY RANGE (column) for range partitions [OK]

Common Mistakes:

• Using PARTITION ON instead of PARTITION BY
• Mixing partition types (LIST or HASH) when RANGE is needed
• Omitting parentheses around column name

3. Given the following commands, what will be the result of querying SELECT * FROM sales WHERE sale_year = 2023;?

CREATE TABLE sales (
  id INT,
  sale_year INT,
  amount NUMERIC
) PARTITION BY LIST (sale_year);

CREATE TABLE sales_2022 PARTITION OF sales FOR VALUES IN (2022);
CREATE TABLE sales_2023 PARTITION OF sales FOR VALUES IN (2023);

INSERT INTO sales VALUES (1, 2022, 100), (2, 2023, 200), (3, 2023, 300);

medium

A. Rows with id 2 and 3 will be returned

B. Rows with id 1 and 2 will be returned

C. Only row with id 1 will be returned

D. No rows will be returned

Solution

1. Step 1: Understand partitioning by LIST on sale_year

   Table sales is partitioned by sale_year with partitions for 2022 and 2023.

2. Step 2: Analyze inserted data and query

   Rows with sale_year 2023 have ids 2 and 3. Query filters sale_year = 2023, so these rows are returned.

3. Final Answer:

   Rows with id 2 and 3 will be returned -> Option A

4. Quick Check:

   Query filters sale_year=2023, returns matching rows [OK]

Hint: Query returns rows matching partition values [OK]

Common Mistakes:

• Assuming all rows return regardless of partition
• Confusing partition column with other columns
• Forgetting to insert data into partitions

4. Identify the error in the following partition creation commands:

CREATE TABLE logs (
  id SERIAL,
  log_date DATE
) PARTITION BY RANGE (log_date);

CREATE TABLE logs_2023 PARTITION OF logs FOR VALUES FROM ('2023-01-01') TO ('2024-01-01');

medium

A. Partitioning by RANGE is not allowed on DATE columns

B. Missing PRIMARY KEY on the parent table

C. The TO value should be '2024-01-01' to include all 2023 dates

D. FOR VALUES clause should use LIST instead of RANGE

Solution

1. Step 1: Check RANGE partition boundaries

   RANGE partitions include values from FROM (inclusive) up to TO (exclusive). To cover all 2023 dates, TO must be '2024-01-01'.

2. Step 2: Analyze given TO value

   TO is '2023-12-31', which excludes that date and any after. This causes missing data for 2023-12-31.

3. Final Answer:

   The TO value should be '2024-01-01' to include all 2023 dates -> Option C

4. Quick Check:

   RANGE TO is exclusive, so use next day [OK]

Hint: RANGE TO value is exclusive; use next day after range end [OK]

Common Mistakes:

• Using inclusive TO value in RANGE partitions
• Thinking RANGE partitioning disallows DATE columns
• Confusing LIST and RANGE partition syntax

5. You want to create a partitioned table events partitioned by HASH on user_id with 4 partitions. Which set of commands correctly creates the table and its partitions?

hard

A. CREATE TABLE events (id INT, user_id INT) PARTITION BY HASH (user_id);
CREATE TABLE events_p0 PARTITION OF events FOR VALUES IN (0);
CREATE TABLE events_p1 PARTITION OF events FOR VALUES IN (1);
CREATE TABLE events_p2 PARTITION OF events FOR VALUES IN (2);
CREATE TABLE events_p3 PARTITION OF events FOR VALUES IN (3);

B. CREATE TABLE events (id INT, user_id INT) PARTITION BY HASH (user_id);
CREATE TABLE events_p0 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 0);
CREATE TABLE events_p1 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 1);
CREATE TABLE events_p2 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 2);
CREATE TABLE events_p3 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 3);

C. CREATE TABLE events (id INT, user_id INT) PARTITION BY LIST (user_id);
CREATE TABLE events_p0 PARTITION OF events FOR VALUES IN (0);
CREATE TABLE events_p1 PARTITION OF events FOR VALUES IN (1);
CREATE TABLE events_p2 PARTITION OF events FOR VALUES IN (2);
CREATE TABLE events_p3 PARTITION OF events FOR VALUES IN (3);

D. CREATE TABLE events (id INT, user_id INT) PARTITION BY RANGE (user_id);
CREATE TABLE events_p0 PARTITION OF events FOR VALUES FROM (0) TO (1);
CREATE TABLE events_p1 PARTITION OF events FOR VALUES FROM (1) TO (2);
CREATE TABLE events_p2 PARTITION OF events FOR VALUES FROM (2) TO (3);
CREATE TABLE events_p3 PARTITION OF events FOR VALUES FROM (3) TO (4);

Solution

1. Step 1: Understand HASH partition syntax

   HASH partitions require FOR VALUES WITH (MODULUS n, REMAINDER r) to define partitions.

2. Step 2: Check each option

   CREATE TABLE events (id INT, user_id INT) PARTITION BY HASH (user_id); CREATE TABLE events_p0 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 0); CREATE TABLE events_p1 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 1); CREATE TABLE events_p2 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 2); CREATE TABLE events_p3 PARTITION OF events FOR VALUES WITH (MODULUS 4, REMAINDER 3); correctly uses HASH partitioning with modulus 4 and remainders 0 to 3. CREATE TABLE events (id INT, user_id INT) PARTITION BY HASH (user_id); CREATE TABLE events_p0 PARTITION OF events FOR VALUES IN (0); CREATE TABLE events_p1 PARTITION OF events FOR VALUES IN (1); CREATE TABLE events_p2 PARTITION OF events FOR VALUES IN (2); CREATE TABLE events_p3 PARTITION OF events FOR VALUES IN (3); uses LIST syntax incorrectly. CREATE TABLE events (id INT, user_id INT) PARTITION BY LIST (user_id); CREATE TABLE events_p0 PARTITION OF events FOR VALUES IN (0); CREATE TABLE events_p1 PARTITION OF events FOR VALUES IN (1); CREATE TABLE events_p2 PARTITION OF events FOR VALUES IN (2); CREATE TABLE events_p3 PARTITION OF events FOR VALUES IN (3); uses LIST partitioning, not HASH. CREATE TABLE events (id INT, user_id INT) PARTITION BY RANGE (user_id); CREATE TABLE events_p0 PARTITION OF events FOR VALUES FROM (0) TO (1); CREATE TABLE events_p1 PARTITION OF events FOR VALUES FROM (1) TO (2); CREATE TABLE events_p2 PARTITION OF events FOR VALUES FROM (2) TO (3); CREATE TABLE events_p3 PARTITION OF events FOR VALUES FROM (3) TO (4); uses RANGE partitioning, not HASH.

3. Final Answer:

   The commands using PARTITION BY HASH (user_id) with FOR VALUES WITH (MODULUS 4, REMAINDER 0-3) -> Option B

4. Quick Check:

   HASH partitions use MODULUS and REMAINDER [OK]

Hint: HASH partitions use MODULUS and REMAINDER in FOR VALUES WITH clause [OK]

Common Mistakes:

• Using FOR VALUES IN instead of FOR VALUES WITH for HASH
• Mixing partition types (LIST or RANGE) with HASH
• Omitting modulus or remainder values