In modern distributed system design, one of the biggest challenges is evenly distributing data across multiple servers and ensuring minimal data movement when nodes are added or removed. This is where consistent hashing becomes critical. But before we dive into consistent hashing, it’s important to understand what hashing is and why it matters.

What is Hashing?

Hashing is a process of converting input data (like a string or number) into a fixed-size numeric value, usually called a hash code or hash value. This is achieved using a hash function.

📌 Hash Function:

A hash function takes a key (e.g., “user123”) and returns an integer — for example:

hash("user123") = 23984719

Then, this hash value is used to determine where to store or retrieve the data.

Why Use Hashing in Systems?

Hashing is widely used in:

  • Databases (e.g., indexing keys)
  • Distributed caches (e.g., Memcached, Redis)
  • Load balancing
  • Sharding in databases
  • Hash tables / maps

It enables fast data access, consistent placement, and efficient key lookups.

Simple Hashing in a Distributed System

In a basic setup, hashing can be used to assign a key to a server like this:

In a basic setup, hashing can be used to assign a key to a server like this:

This ensures that each key maps to a server. But here’s the problem: what happens if the number of servers changes?

Problem with Traditional Hashing

Let’s say we start with 4 servers:

serverIndex = hash(key) % 4;

Now if a 5th server is added, the formula becomes:

serverIndex = hash(key) % 5;

This small change causes most keys to be remapped to new servers, leading to:

  • Data reshuffling
  • Cache misses
  • Increased network load
  • Downtime or lag during scaling

This is where consistent hashing shines.

What is Consistent Hashing?

Consistent Hashing is a smarter way to distribute keys across servers such that only a small subset of keys need to be reassigned when servers are added or removed.

How Consistent Hashing Works

Step-by-step Breakdown:

  1. Hash Ring: Imagine a circle (a ring) where both keys and servers are hashed onto it.
  2. Key Placement: Each key is assigned to the first server in a clockwise direction from the key’s position on the ring.
  3. Dynamic Scaling: When a server joins or leaves, only its neighboring keys get remapped — not all keys.

Example:

  • Servers: A, B, C hashed to points 100, 200, 300 on the ring.
  • A key hashing to 150 will go to Server B (next clockwise).
  • If Server D joins and hashes to 180, only keys between 150–180 move to D. Others remain unaffected.

Virtual Nodes: Enhancing Load Distribution

To improve key distribution, each server is assigned multiple virtual nodes:

  • Server A → Virtual nodes at 100, 140, 180
  • Server B → Virtual nodes at 210, 250, 290

This ensures better balance and reduces hot spots.

Advantages of Consistent Hashing

  • Low key remapping when nodes change
  • Even load distribution with virtual nodes
  • Efficient scaling in distributed systems
  • Supports high availability and fault tolerance

Where is Consistent Hashing Used?

  • Amazon DynamoDB – Data sharding
  • Cassandra – Token ring-based partitioning
  • Redis Cluster
  • Akamai CDN – Routing client requests
  • Consistent load balancing across microservices

Limitations:

  • Slightly complex to implement
  • Requires robust failure detection
  • Needs replication strategies for redundancy

Conclusion: Why Learn Consistent Hashing

If you’re preparing for system design interviews or building resilient distributed applications, consistent hashing is a must-know pattern. It’s used by giants like Amazon, Google, and Netflix to handle millions of users and data points efficiently.

By combining the principles of hashing with intelligent node mapping, consistent hashing makes systems highly scalable, fault-tolerant, and efficient.

References:

Other HLD Topics to look at:

Interview Experiences:

Categorized in:

HLD, Must Know HLD,