How to design consistent hashing

Introduction - What is Hashing ?

A hash function is a function that maps one piece of data—typically describing some kind of object, often of arbitrary size—to another piece of data, typically an integer, known as hash code, or simply hash.

Hashing is widely used in algorithms, data structures, and cryptography.

1. Hash Functions
   Hash functions take variable-length input data and produce a fixed-length output value.
   

   public int hashCode() {
       int h = hash;
       if (h == 0 && value.length > 0) {
           char val[] = value;
           for (int i = 0; i < value.length; i++) {
               h = 31 * h + val[i];
           }
           hash = h;
       }
       return h;
   }
   
   public static void main(String[] args) {
       String anotherTest = "java";
       String test = "test";
       String oneMoreTest = "dev";
       System.out.println(test.hashCode()); // output: 3556498
       System.out.println(test.hashCode()); // output: 3556498
       System.out.println(anotherTest.hashCode()); // output: 3254818
       System.out.println(oneMoreTest.hashCode()); // output: 99349
   }

2. Hashing Algo
   Hashing is a technique or process of mapping keys, values into the hash table by using a hash function : Data Structure and Algorithms - Hash Table (tutorialspoint.com)
   

3. Cryptographic Hash Functions
   Cryptographic hash functions are a specialized group of hash functions. They provide an increased level of security. Thus, they are used for cryptography purposes like password verification, data integrity validation, blockchain (cryptocurrencies).

The Rehashing problem : How to balance the load to x servers ?

Why do we need Hashing first ?

But, hashing is a common way to balance the load to servers : serverIndex = hash(key) % N where N is the size of the server pool. (See an example in the page : the hashes for 4 servers - Hash % 4 servers).

But, the problem is when the servers pool is not static anymore ! New servers are added and existing servers could be removed.

If we removed 1 server, modulo changes : Hash % 3 servers. Most of keys are redistributed.

But, when 1 server goes offline, most cache clients will connect to the wrong servers to fetch data. We have an horizontal scalability problem.

The solution : Consistent hashing (CH)

How does CH work ?

CH facilitates the distribution of data across a set of nodes (DB servers) in such a way that minimizes the mapping or reorganization of data when nodes are added or removed.

1- What is CH ? Creating th Hash Key Space.

We need to consider a hash function that generates integer hash values in the following range [0, 2^32-1] and the output of a hash function is: x0, x1, x2,…,xN.

2- Representing the hash space as a hash ring

By connecting both ends, we get a hash ring. So, we can visualize it as a ring.

3- Hash servers : Placing DB servers in the hash ring

We can use the hash function and so, hash the servers on their IP address to map them to different integers.

4- Hash Keys: Determining placement of keys on servers

To find which DB server an incoming key resides on : we assume we have 4 incoming keys (key0, key1, key2, key3) and none of them directly maps the hash value of any 4 servers on our ring. We follow the hand of a clock and insert the key.

5- Adding a server in the ring

When we add a server, we need to add it between server 0 and server 3. We'll need to remap only to remap the keys to the server 4 (new server).

So, we need to remap only k/n keys where k is the number of keys (4) and n is the number of servers (5).

6- Removing a server from the ring