Data replication strategies in HLD - System Design Guide

Problem Statement

When a single database instance handles all read and write requests, it becomes a bottleneck causing slow response times and risking data loss if the server fails. Without replication, data availability and fault tolerance are severely limited, leading to downtime and poor user experience.

Solution

Data replication copies data across multiple servers to distribute load and increase availability. Writes are propagated to replicas either synchronously or asynchronously, allowing reads to be served from multiple nodes and providing fault tolerance if one node fails.

Architecture

Primary DB

Primary DB

↓

Replica 1

This diagram shows a primary database replicating data to two replicas. Clients send writes to the primary and reads to replicas to balance load and improve availability.

Trade-offs

✓ Pros

→

Improves read scalability by distributing read traffic across replicas.

→

Increases fault tolerance by providing data copies if the primary fails.

→

Reduces latency for geographically distributed users by placing replicas closer to them.

✗ Cons

→

Synchronous replication can increase write latency due to waiting for replicas to confirm.

→

Asynchronous replication risks data loss if the primary fails before replicas catch up.

→

Complexity in handling data consistency and conflict resolution increases.

Use when read traffic is significantly higher than write traffic (e.g., 10x reads), or when high availability and disaster recovery are required for critical data.

Avoid when write latency must be minimal and strict consistency is required, or when system scale is small (under 1000 writes per second) where replication overhead outweighs benefits.

Real World Examples

Amazon

Uses asynchronous replication across multiple regions to ensure data durability and low-latency reads for global customers.

Netflix

Employs multi-region replication to serve streaming metadata with low latency and high availability.

Google

Uses synchronous replication in Spanner to provide strong consistency across distributed nodes.

Alternatives

Sharding

Splits data horizontally across multiple databases instead of copying full data sets.

Use when: Use when write throughput is very high and data can be partitioned by key.

Caching

Stores frequently accessed data temporarily in fast storage instead of replicating full data sets.

Use when: Use when read latency is critical and data freshness can be relaxed.

Summary

Data replication copies data across multiple servers to improve availability and read scalability.

It balances trade-offs between consistency, latency, and fault tolerance depending on replication type.

Choosing the right replication strategy depends on system scale, read/write ratio, and consistency needs.

Practice

1. Which data replication strategy ensures that all copies of data are updated at the same time before confirming a write operation?

easy

A. Synchronous replication

B. Asynchronous replication

C. Master-slave replication

D. Multi-master replication

Solution

1. Step 1: Understand synchronous replication

   Synchronous replication waits for all copies to confirm the write before completing the operation.

2. Step 2: Compare with other strategies

   Asynchronous replication does not wait, master-slave and multi-master describe topology, not timing.

3. Final Answer:

   Synchronous replication -> Option A

4. Quick Check:

   Strong consistency = Synchronous replication [OK]

Hint: Strong consistency means waiting for all copies to update [OK]

Common Mistakes:

• Confusing asynchronous with synchronous replication
• Mixing replication timing with replication topology
• Assuming master-slave means synchronous

2. Which of the following is the correct description of asynchronous replication?

easy

A. Replicas confirm the data before the primary confirms the write

B. Data is copied to replicas without waiting for confirmation from them

C. Data is copied simultaneously to all replicas with confirmation

D. Data is copied to replicas before the primary confirms the write

Solution

1. Step 1: Define asynchronous replication

   It copies data to replicas but does not wait for their confirmation before confirming the write.

2. Step 2: Eliminate other options

   Replicas confirm the data before the primary confirms the write requires replica confirmation before primary ack (synchronous); B copies before primary ack (incorrect); C describes synchronous behavior.

3. Final Answer:

   Data is copied to replicas without waiting for confirmation from them -> Option B

4. Quick Check:

   Asynchronous = no wait for replica confirmation [OK]

Hint: Asynchronous means no waiting for replicas to confirm [OK]

Common Mistakes:

• Thinking asynchronous waits for replicas
• Confusing timing of data copy
• Mixing synchronous and asynchronous definitions

3. Consider a system using master-slave replication. If the master receives a write request, what happens next?

medium

A. The write is applied simultaneously to master and slaves synchronously

B. The write is applied to all slaves first, then to the master

C. The write is only applied to slaves, not the master

D. The write is applied to the master and then asynchronously copied to slaves

Solution

1. Step 1: Understand master-slave replication flow

   Writes go to the master first, then are copied to slaves, usually asynchronously.

2. Step 2: Check other options

   Slaves do not receive writes first or alone; synchronous writes to all are not typical in master-slave.

3. Final Answer:

   The write is applied to the master and then asynchronously copied to slaves -> Option D

4. Quick Check:

   Master handles writes first in master-slave [OK]

Hint: Master handles writes first, slaves get updates later [OK]

Common Mistakes:

• Assuming slaves can accept writes first
• Confusing synchronous with asynchronous replication
• Thinking slaves are independent masters

4. A system using multi-master replication faces conflicts when two masters update the same data simultaneously. What is a common way to fix this issue?

medium

A. Switch to asynchronous replication only

B. Disable writes on all but one master

C. Use conflict resolution strategies like timestamps or version vectors

D. Remove all replicas and use a single database

Solution

1. Step 1: Identify conflict in multi-master replication

   Simultaneous writes cause conflicts that need resolution strategies.

2. Step 2: Evaluate conflict resolution methods

   Timestamps or version vectors help decide which write wins or how to merge.

3. Final Answer:

   Use conflict resolution strategies like timestamps or version vectors -> Option C

4. Quick Check:

   Conflict resolution = timestamps/version vectors [OK]

Hint: Resolve conflicts with timestamps or version vectors [OK]

Common Mistakes:

• Disabling multi-master writes defeats purpose
• Thinking asynchronous replication solves conflicts
• Removing replicas is impractical

5. You are designing a global application with users worldwide needing fast reads and writes. Which replication strategy best fits this use case?

hard

A. Multi-master replication with conflict resolution

B. Master-slave replication with synchronous writes

C. Single master with asynchronous replication

D. No replication, single database instance

Solution

1. Step 1: Analyze global user needs

   Users worldwide require fast reads and writes from multiple locations.

2. Step 2: Match replication strategy

   Multi-master allows writes at multiple locations with conflict resolution, improving speed and availability.

3. Step 3: Eliminate other options

   Master-slave with synchronous writes is slow globally; single master limits write locations; no replication is slow and risky.

4. Final Answer:

   Multi-master replication with conflict resolution -> Option A

5. Quick Check:

   Global fast writes = multi-master + conflict resolution [OK]

Hint: Global writes need multi-master with conflict handling [OK]

Common Mistakes:

• Choosing synchronous master-slave for global writes
• Ignoring conflict resolution in multi-master
• Using single master limits write speed globally