How horizontal scaling differs from vertical scaling

Short answer

Horizontal scaling adds more machines to distribute load, while vertical scaling increases the power of existing machines.

Key differences

• Cost: Horizontal scaling uses commodity hardware; vertical scaling requires expensive high-end servers.
• Limit: Vertical scaling hits physical hardware limits; horizontal scaling can grow theoretically without bound.
• Complexity: Horizontal scaling requires distributed system techniques; vertical scaling is simpler but less resilient.
• Elasticity: Horizontal scaling supports auto-scaling; vertical scaling requires downtime to resize.
• Fault tolerance: Horizontal scaling offers natural redundancy; vertical scaling creates a single point of failure.

When to use each

• Use vertical scaling for monolithic databases or stateful services that are hard to partition.
• Use horizontal scaling for stateless web servers and distributed data stores.
• Combine both by scaling up database primaries and scaling out read replicas.
• Start vertical and move horizontal when cost or uptime requirements demand it.

Example

# Consistent hashing for service discovery
import hashlib

def get_node(key, nodes):
    hash_val = int(hashlib.md5(key.encode()).hexdigest(), 16)
    return nodes[hash_val % len(nodes)]

node = get_node('user-123', ['node-a', 'node-b', 'node-c'])

This snippet implements consistent hashing to distribute keys across nodes, a foundational technique in scalable distributed systems.

Additional context

Applying these principles consistently across projects leads to more maintainable systems, clearer team communication, and better outcomes for end users. Regular review and refinement of practices ensure continuous improvement.