Why clustering provides horizontal scaling in Redis - Performance Analysis

Time Complexity: Why clustering provides horizontal scaling

O(n / k)

Understanding Time Complexity

We want to see how adding more servers in a Redis cluster affects the work done when handling data.

How does splitting data across servers change the time to process requests?

Scenario Under Consideration

Analyze the time complexity of these Redis cluster commands.

CLUSTER INFO
SET key1 value1
SET key2 value2
GET key1
GET key2
    

This code shows simple commands where keys are stored and retrieved in a Redis cluster.

Identify Repeating Operations

Look at what repeats when we add more keys and servers.

• Primary operation: Storing or retrieving a key on the correct server.
• How many times: Once per key, but keys are spread across servers.

How Execution Grows With Input

As we add more keys, each server handles fewer keys because data is split.

Input Size (n keys)	Approx. Operations per Server
10 keys, 1 server	10 operations
100 keys, 5 servers	~20 operations each
1000 keys, 10 servers	~100 operations each

Pattern observation: Adding servers splits the work, so each server does less work even if total keys grow.

Final Time Complexity

Time Complexity: O(n / k)

This means the work per server grows with the number of keys divided by the number of servers.

Common Mistake

[X] Wrong: "Adding more servers always makes the system twice as fast regardless of data size."

[OK] Correct: Because if data is small, splitting it adds overhead. Also, network and coordination costs affect speed.

Interview Connect

Understanding how clustering spreads work helps you explain scaling in real systems. It shows you can think about how adding resources changes performance.

Self-Check

What if we added more servers but kept the same number of keys? How would the time complexity per server change?

Practice

1. Why does Redis clustering provide horizontal scaling?

easy

A. Because it makes a single machine faster

B. Because it uses a single database instance

C. Because it compresses data to save space

D. Because it adds more machines to share the load

Solution

1. Step 1: Understand horizontal scaling meaning

   Horizontal scaling means adding more machines to handle more work, not just improving one machine.

2. Step 2: Connect clustering to horizontal scaling

   Redis clustering splits data across many machines, so the system can handle more users and data by adding machines.

3. Final Answer:

   Because it adds more machines to share the load -> Option D

4. Quick Check:

   Horizontal scaling = adding machines [OK]

Hint: Horizontal scaling means adding machines, not speeding one [OK]

Common Mistakes:

• Thinking clustering speeds up a single machine
• Confusing horizontal with vertical scaling
• Believing clustering compresses data

2. Which Redis command is used to create a cluster for horizontal scaling?

easy

A. CLUSTER MEET

B. SET

C. GET

D. FLUSHALL

Solution

1. Step 1: Identify cluster creation commands

   Redis uses special cluster commands like CLUSTER MEET to connect nodes and form a cluster.

2. Step 2: Check other commands' purpose

   SET and GET are for data operations, FLUSHALL clears data, not for clustering.

3. Final Answer:

   CLUSTER MEET -> Option A

4. Quick Check:

   Cluster setup uses CLUSTER MEET [OK]

Hint: Cluster commands start with CLUSTER keyword [OK]

Common Mistakes:

• Choosing data commands like SET or GET
• Confusing FLUSHALL with cluster setup
• Not knowing cluster-specific commands

3. Given this Redis cluster setup code snippet, what happens when you add a new node?

CLUSTER MEET 192.168.1.2 7001
CLUSTER MEET 192.168.1.3 7001

medium

A. The new node replaces an existing node

B. The new node joins the cluster and shares data load

C. The new node disconnects the cluster

D. The new node deletes all data

Solution

1. Step 1: Understand CLUSTER MEET command effect

   CLUSTER MEET connects a new node to the cluster, allowing it to join and share data.

2. Step 2: Check other options' validity

   It does not replace nodes, disconnect cluster, or delete data by itself.

3. Final Answer:

   The new node joins the cluster and shares data load -> Option B

4. Quick Check:

   CLUSTER MEET adds node to cluster [OK]

Hint: CLUSTER MEET adds nodes, expanding cluster horizontally [OK]

Common Mistakes:

• Thinking new node replaces old nodes
• Assuming cluster disconnects on adding nodes
• Believing data is deleted on node join

4. You tried to add a node to a Redis cluster using CLUSTER MEET, but the node does not join. What is the likely error?

medium

A. The node IP address is unreachable

B. The command syntax is incorrect

C. The new node has a different cluster configuration

D. The cluster is already full and cannot add nodes

Solution

1. Step 1: Check network connectivity

   If the new node IP is unreachable, it cannot join the cluster despite correct commands.

2. Step 2: Validate other options

   Different config or full cluster usually give errors; syntax errors cause command failure, not silent join failure.

3. Final Answer:

   The node IP address is unreachable -> Option A

4. Quick Check:

   Unreachable IP blocks cluster join [OK]

Hint: Check network reachability if node won't join cluster [OK]

Common Mistakes:

• Ignoring network issues
• Assuming syntax error without checking
• Believing cluster size limits node addition

5. You have a Redis cluster with 3 nodes. To handle more users, you add 2 more nodes. How does this improve horizontal scaling?

hard

A. New nodes only store backups, no load increase

B. All data moves to the new nodes, old nodes become backups

C. Data and requests are split across 5 nodes, increasing capacity

D. The cluster merges all nodes into one big node

Solution

1. Step 1: Understand data distribution in clustering

   Redis cluster splits data and requests across all nodes to balance load.

2. Step 2: Analyze effect of adding nodes

   Adding nodes increases total capacity by sharing data and requests among more machines.

3. Step 3: Eliminate wrong options

   Data does not move only to new nodes, cluster does not merge nodes, and new nodes are not just backups.

4. Final Answer:

   Data and requests are split across 5 nodes, increasing capacity -> Option C

5. Quick Check:

   More nodes = more shared load = horizontal scaling [OK]

Hint: More nodes share data and requests, boosting capacity [OK]

Common Mistakes:

• Thinking new nodes only backup data
• Believing cluster merges nodes into one
• Assuming data moves only to new nodes