RDB vs AOF comparison in Redis

Introduction

RDB and AOF are two ways Redis saves data to keep it safe. They help recover data if Redis stops or crashes.

You want fast recovery with less disk space use.

You need to save data very often to avoid losing recent changes.

You want a balance between performance and data safety.

You want to understand how Redis saves data to choose the best method.

You want to troubleshoot or optimize Redis persistence.

Syntax

Redis

RDB: Saves snapshots of data at intervals.
AOF: Logs every write operation to a file.

RDB creates a point-in-time snapshot of the database.

AOF records each command that changes the data, like a diary.

Examples

RDB example: Save snapshot if 1 change in 900 seconds, or 10 changes in 300 seconds, etc.

Redis

save 900 1
save 300 10
save 60 10000

AOF example: Turn on AOF and sync to disk every second for safety.

Redis

appendonly yes
appendfsync everysec

Sample Program

This config saves snapshots at intervals and also logs every write command every second.

Redis

# This is a Redis config snippet showing RDB and AOF settings
save 900 1
save 300 10
save 60 10000
appendonly yes
appendfsync everysec

Important Notes

RDB is faster for backups but may lose recent changes if Redis crashes.

AOF is safer for data but can be slower and use more disk space.

You can use both together for better safety and performance.

Summary

RDB saves snapshots at intervals, good for fast recovery.

AOF logs every change, good for data safety.

Choosing depends on your needs for speed and data safety.

Practice

1. Which Redis persistence method saves the entire dataset at specific time intervals?

easy

A. AOF (Append Only File)

B. RDB (Redis Database snapshot)

C. Both RDB and AOF save snapshots

D. Neither RDB nor AOF save snapshots

Solution

1. Step 1: Understand RDB persistence

   RDB saves the dataset by creating snapshots at set intervals.

2. Step 2: Compare with AOF

   AOF logs every write operation, not snapshots.

3. Final Answer:

   RDB (Redis Database snapshot) -> Option B

4. Quick Check:

   Snapshot saving = RDB [OK]

Hint: RDB = snapshots at intervals, AOF = logs every change [OK]

Common Mistakes:

• Confusing AOF with snapshot saving
• Thinking both save snapshots
• Believing neither saves snapshots

2. Which of the following is the correct Redis command to enable AOF persistence?

easy

A. CONFIG SET appendonly yes

B. CONFIG SET save "900 1 300 10 60 10000"

C. CONFIG SET rdbcompression no

D. CONFIG SET maxmemory-policy allkeys-lru

Solution

1. Step 1: Identify AOF enabling command

   The command to enable AOF is CONFIG SET appendonly yes.

2. Step 2: Check other options

   CONFIG SET save "900 1 300 10 60 10000" sets RDB save intervals, C disables RDB compression, D sets memory eviction policy.

3. Final Answer:

   CONFIG SET appendonly yes -> Option A

4. Quick Check:

   Enable AOF = appendonly yes [OK]

Hint: Enable AOF with 'appendonly yes' command [OK]

Common Mistakes:

• Confusing RDB save command with AOF
• Using unrelated config commands
• Forgetting to set appendonly to yes

3. Given Redis configured with AOF enabled and fsync set to 'everysec', what is the expected data durability behavior?

medium

A. Data is saved only at intervals, risking data loss of last seconds

B. Every write is immediately saved to disk, no data loss

C. Data is never saved to disk automatically

D. Data is saved only when Redis restarts

Solution

1. Step 1: Understand AOF with fsync everysec

   With fsync set to 'everysec', Redis writes AOF data to disk every second.

2. Step 2: Analyze durability implications

   This means data could be lost if Redis crashes within the last second before fsync.

3. Final Answer:

   Data is saved only at intervals, risking data loss of last seconds -> Option A

4. Quick Check:

   AOF everysec fsync = possible 1-second data loss [OK]

Hint: AOF everysec fsync means possible 1-second data loss [OK]

Common Mistakes:

• Assuming every write is immediately saved
• Thinking data is never saved automatically
• Confusing with no persistence

4. You enabled both RDB and AOF persistence in Redis, but after a crash, data is missing. What is the most likely cause?

medium

A. Redis does not support using both RDB and AOF together

B. Redis was not restarted after enabling persistence

C. RDB snapshot interval is too long, losing recent changes

D. AOF file is corrupted and Redis cannot load it

Solution

1. Step 1: Understand RDB and AOF coexistence

   Redis supports both RDB and AOF, preferring to load AOF first on restart if available.

2. Step 2: Analyze data loss causes

   With both enabled, if data is missing after a crash, the most likely cause is AOF corruption, causing fallback to the older RDB snapshot and loss of recent changes.

3. Final Answer:

   AOF file is corrupted and Redis cannot load it -> Option D

4. Quick Check:

   Data loss with both = AOF corrupted [OK]

Hint: With both enabled, data loss likely from AOF corruption [OK]

Common Mistakes:

• Believing Redis can't use both persistences
• Thinking RDB interval too long is the cause
• Forgetting to restart Redis

5. You want Redis to recover data quickly after a crash but can tolerate losing up to 1 minute of recent writes. Which persistence setup is best?

hard

A. Enable AOF with fsync always, disable RDB

B. Enable both RDB and AOF with fsync everysec

C. Enable RDB snapshots every 60 seconds, disable AOF

D. Disable both RDB and AOF for fastest performance

Solution

1. Step 1: Match recovery speed and data loss tolerance

   RDB snapshots every 60 seconds allow fast recovery but may lose up to 1 minute of data.

2. Step 2: Compare with AOF options

   AOF with fsync always is slower but safer; both enabled adds overhead; disabling persistence risks total data loss.

3. Final Answer:

   Enable RDB snapshots every 60 seconds, disable AOF -> Option C

4. Quick Check:

   Fast recovery + 1 min loss = RDB every 60s [OK]

Hint: RDB snapshots match fast recovery with some data loss [OK]

Common Mistakes:

• Choosing AOF for fast recovery despite slower startup
• Enabling both causing unnecessary overhead
• Disabling persistence risking total data loss