Cross-datacenter replication in Kafka - Step-by-Step Execution

Process Flow - Cross-datacenter replication

Produce message in DC1

↓

Message stored in DC1 Kafka cluster

↓

Replication process reads message

↓

Send message over network to DC2

↓

Message stored in DC2 Kafka cluster

↓

Consumer reads message from DC2

Messages produced in one data center are stored locally, then replicated over the network to another data center's Kafka cluster for consumption.

Execution Sample

Kafka

producer.send('topic', 'message1')
# Message stored in DC1
replicator.replicate('topic')
# Message sent to DC2
consumer.read('topic')
# Consumer reads replicated message

This code simulates producing a message in one data center, replicating it to another, and consuming it there.

Process Table

Step	Action	Source DC	Message State	Replication Status	Consumer Output
1	Produce message 'message1'	DC1	Stored in DC1	Not started	None
2	Start replication	DC1	Stored in DC1	Sending to DC2	None
3	Message received in DC2	DC2	Stored in DC2	Completed	None
4	Consumer reads message	DC2	Stored in DC2	Completed	'message1' read
5	No new messages	DC1 & DC2	No change	Idle	No output

💡 Replication completes after message is stored in DC2 and consumed.

Status Tracker

Variable	Start	After Step 1	After Step 2	After Step 3	After Step 4	Final
message_storage_DC1	empty	message1	message1	message1	message1	message1
message_storage_DC2	empty	empty	empty	message1	message1	message1
replication_status	idle	idle	sending	completed	completed	idle
consumer_output	none	none	none	none	message1 read	none

Key Moments - 3 Insights

Why does the message appear in DC2 only after replication starts?

Because the message is first stored locally in DC1 (see Step 1), and only after replication begins (Step 2) is it sent and stored in DC2 (Step 3).

Can the consumer read the message before replication completes?

No, the consumer in DC2 reads the message only after it is stored there (Step 4), so before replication completes, there is no message to read.

What happens if no new messages are produced?

Replication stays idle (Step 5), and consumers have no new messages to read, so no output occurs.

Visual Quiz - 3 Questions

Look at the execution table, what is the replication status at Step 3?

A Completed

B Sending

C Idle

D Not started

💡 Hint

Check the 'Replication Status' column at Step 3 in the execution_table.

At which step does the consumer read the replicated message?

A Step 2

B Step 3

C Step 4

D Step 5

💡 Hint

Look at the 'Consumer Output' column in the execution_table to find when 'message1 read' appears.

If the producer sends a second message before replication starts, what changes in the variable_tracker?

A message_storage_DC2 will have two messages immediately

B message_storage_DC1 will have two messages after Step 1

C replication_status will be 'completed' at Step 1

D consumer_output will show both messages read at Step 2

💡 Hint

Check how message_storage_DC1 changes after Step 1 in variable_tracker.

Concept Snapshot

Cross-datacenter replication in Kafka:
- Messages produced in one data center (DC1) are stored locally.
- A replication process sends messages over the network to another data center (DC2).
- Messages are stored in DC2's Kafka cluster.
- Consumers in DC2 read replicated messages.
- Replication ensures data availability across locations.

Full Transcript

Cross-datacenter replication means sending Kafka messages from one data center to another. First, a message is produced and stored in the first data center's Kafka cluster. Then, a replication process reads this message and sends it over the network to the second data center. The message is stored there, and consumers in the second data center can read it. This process ensures that data is available in multiple locations for reliability and faster access. The execution table shows each step: producing, replicating, storing, and consuming. Variables track message storage and replication status. Key moments clarify when messages appear and can be read. The quiz tests understanding of replication status and message flow.