Logical replication basics in PostgreSQL - Step-by-Step Execution

Concept Flow - Logical replication basics

Setup Publisher

↓

Create Publication

↓

Setup Subscriber

↓

Create Subscription

↓

Data Changes on Publisher

↓

Changes Sent as WAL Logs

↓

Subscriber Applies Changes

↓

Data Synced on Subscriber

Logical replication starts by setting up a publisher and subscriber. Changes on the publisher are sent as logs and applied on the subscriber to keep data in sync.

Execution Sample

PostgreSQL

CREATE PUBLICATION mypub FOR TABLE users;
CREATE SUBSCRIPTION mysub CONNECTION 'host=pubhost dbname=mydb user=replicator password=secret' PUBLICATION mypub;

This code creates a publication on the publisher for the 'users' table and a subscription on the subscriber to receive changes.

Execution Table

Step	Action	Details	Result
1	Create Publication	CREATE PUBLICATION mypub FOR TABLE users;	Publication 'mypub' created on publisher
2	Create Subscription	CREATE SUBSCRIPTION mysub CONNECTION 'host=pubhost dbname=mydb user=replicator password=secret' PUBLICATION mypub;	Subscription 'mysub' created on subscriber
3	Insert Data on Publisher	INSERT INTO users VALUES (1, 'Alice');	Change logged in WAL
4	Send WAL Logs	Logical replication sends changes from publisher to subscriber	Change sent to subscriber
5	Apply Changes on Subscriber	Subscriber applies insert to 'users' table	Data inserted on subscriber
6	Verify Data	SELECT * FROM users;	Row (1, 'Alice') present on subscriber
7	Exit	Replication running continuously	Process ongoing, no exit

💡 Logical replication runs continuously to keep data synced; no termination unless stopped

Variable Tracker

Variable	Start	After Step 3	After Step 5	Final
users table on publisher	empty	1 row inserted	1 row inserted	1 row inserted
users table on subscriber	empty	empty	1 row inserted	1 row inserted
WAL logs	empty	insert logged	sent and applied	up to date

Key Moments - 3 Insights

Why does the subscriber not have data before creating the subscription?

Because data is only sent after the subscription is created and connected, as shown in execution_table step 4 and 5.

Does logical replication copy the entire database automatically?

No, it replicates only the tables included in the publication, as seen in step 1 where only 'users' table is published.

What happens if data changes on the publisher after subscription creation?

Changes are continuously sent and applied on the subscriber, keeping data in sync as shown in steps 4 and 5.

Visual Quiz - 3 Questions

Look at the execution_table, what is the state of the 'users' table on the subscriber after step 3?

A Empty

B 1 row inserted

C Subscription created

D Data deleted

💡 Hint

Check variable_tracker column 'After Step 3' for 'users table on subscriber'

At which step does the subscriber apply the data change?

A Step 2

B Step 3

C Step 5

D Step 6

💡 Hint

Look at execution_table row describing 'Apply Changes on Subscriber'

If the publication included two tables, how would the replication change?

A Only one table would replicate

B Both tables' changes would replicate

C No tables would replicate

D Subscriber would reject subscription

💡 Hint

Refer to key_moments about publication scope and step 1 in execution_table

Concept Snapshot

Logical replication basics in PostgreSQL:
- Publisher creates a publication for specific tables
- Subscriber creates a subscription to that publication
- Data changes on publisher are sent as WAL logs
- Subscriber applies changes to keep data synced
- Replication runs continuously until stopped

Full Transcript

Logical replication in PostgreSQL involves setting up a publisher and subscriber. The publisher creates a publication specifying which tables to replicate. The subscriber creates a subscription connecting to the publisher and subscribing to that publication. When data changes occur on the publisher, these changes are sent as write-ahead logs (WAL) to the subscriber. The subscriber applies these changes to its tables, keeping data synchronized. This process runs continuously, ensuring ongoing replication of data changes. The example shows creating a publication for the 'users' table, creating a subscription, inserting data on the publisher, and the data appearing on the subscriber after replication.

Practice

1. What is the main purpose of logical replication in PostgreSQL?

easy

A. To copy data changes from specific tables between databases

B. To create a full backup of the database

C. To optimize query performance

D. To encrypt data during transfer

Solution

1. Step 1: Understand logical replication concept

   Logical replication copies only data changes from selected tables, not the entire database.

2. Step 2: Compare options with concept

   Encrypting data during transfer, creating a full backup of the database, and optimizing query performance are unrelated to logical replication.

3. Final Answer:

   To copy data changes from specific tables between databases -> Option A

4. Quick Check:

   Logical replication = Copy data changes [OK]

Hint: Logical replication copies changes, not full backups [OK]

Common Mistakes:

• Confusing logical replication with physical backup
• Thinking it copies entire database
• Assuming it improves query speed

2. Which SQL command is used to create a publication for logical replication?

easy

A. CREATE REPLICATION SLOT myslot LOGICAL;

B. CREATE SUBSCRIPTION mysub CONNECTION 'conninfo' PUBLICATION mypub;

C. CREATE PUBLICATION mypub FOR ALL TABLES;

D. CREATE DATABASE mydb WITH REPLICATION;

Solution

1. Step 1: Identify command for publication creation

   The command to create a publication is CREATE PUBLICATION followed by publication name and tables.

2. Step 2: Analyze options

   Only CREATE PUBLICATION mypub FOR ALL TABLES; creates a publication. CREATE SUBSCRIPTION mysub CONNECTION 'conninfo' PUBLICATION mypub; creates a subscription, CREATE REPLICATION SLOT myslot LOGICAL; creates a replication slot, and CREATE DATABASE mydb WITH REPLICATION; creates a database.

3. Final Answer:

   CREATE PUBLICATION mypub FOR ALL TABLES; -> Option C

4. Quick Check:

   Publication creation = CREATE PUBLICATION [OK]

Hint: Publication uses CREATE PUBLICATION, subscription uses CREATE SUBSCRIPTION [OK]

Common Mistakes:

• Mixing publication and subscription commands
• Using replication slot command for publication
• Confusing database creation with replication setup

3. Given the following commands on the publisher:

CREATE PUBLICATION mypub FOR TABLE customers;

And on the subscriber:

CREATE SUBSCRIPTION mysub CONNECTION 'host=source dbname=mydb user=replicator password=secret' PUBLICATION mypub;

What will happen when a new row is inserted into the customers table on the publisher?

medium

A. The new row will be replicated to the subscriber's customers table

B. The new row will not be replicated because subscription is missing

C. The entire customers table will be copied again

D. An error will occur because publications cannot replicate inserts

Solution

1. Step 1: Understand publication and subscription setup

   The publication includes the customers table, and the subscription connects to it, enabling replication of changes.

2. Step 2: Analyze effect of insert on replication

   Inserts on the published table are sent to the subscriber, so the new row will appear there.

3. Final Answer:

   The new row will be replicated to the subscriber's customers table -> Option A

4. Quick Check:

   Insert on published table = replicated row [OK]

Hint: Inserts on published tables replicate if subscription exists [OK]

Common Mistakes:

• Thinking subscription is missing when it is created
• Assuming full table copy on each insert
• Believing inserts are not replicated

4. You created a subscription but notice no data is replicating. Which of the following is a likely cause?

medium

A. Logical replication does not support inserts

B. The subscriber database is offline

C. You forgot to create a replication slot on the subscriber

D. The publication does not include the tables you want to replicate

Solution

1. Step 1: Check publication includes tables

   If the publication does not include the desired tables, no changes will be sent to the subscriber.

2. Step 2: Evaluate other options

   The subscriber database is offline is unlikely because subscription requires the subscriber to be online. Logical replication does not support inserts is false; inserts are supported. You forgot to create a replication slot on the subscriber is incorrect because replication slots are created on the publisher, not subscriber.

3. Final Answer:

   The publication does not include the tables you want to replicate -> Option D

4. Quick Check:

   Missing tables in publication = no replication [OK]

Hint: Ensure publication includes tables to replicate [OK]

Common Mistakes:

• Confusing replication slot location
• Assuming inserts are unsupported
• Ignoring publication table list

5. You want to replicate only changes from the orders table but exclude the order_logs table, which is large and not needed on the subscriber. How should you set up the publication?

hard

A. CREATE PUBLICATION mypub FOR TABLE orders, order_logs;

B. CREATE PUBLICATION mypub FOR TABLE orders;

C. CREATE PUBLICATION mypub FOR ALL TABLES;

D. CREATE PUBLICATION mypub FOR TABLE order_logs;

Solution

1. Step 1: Understand selective table replication

   To replicate only the orders table, the publication must include only that table.

2. Step 2: Analyze options for correct table inclusion

   CREATE PUBLICATION mypub FOR ALL TABLES; replicates all tables, including order_logs which is unwanted. CREATE PUBLICATION mypub FOR TABLE orders, order_logs; includes both tables. CREATE PUBLICATION mypub FOR TABLE order_logs; includes only order_logs, which is unwanted.

3. Final Answer:

   CREATE PUBLICATION mypub FOR TABLE orders; -> Option B

4. Quick Check:

   Include only needed tables in publication [OK]

Hint: List only needed tables in publication to exclude others [OK]

Common Mistakes:

• Using FOR ALL TABLES when exclusion is needed
• Including unwanted tables in publication
• Confusing publication and subscription roles