Horizontal scaling, or scaling out, increases performance by distributing load across existing and additional processing resources. Unlike upgrading one processor or adding memory, it grows capacity by adding resources that share the work. section 1.3
The approach addresses several limits of scaling up:
- Adding processing resources can avoid the performance ceiling of increasingly expensive, more powerful individual processors. section 1.3
- Distributed resources support real-time systems, high data volumes, reliability through redundancy, and improved resource utilization. section 1.3
- A single computer may not be enough; multiple interconnected machines can form a computing cluster for data-processing tasks. section 1.3
Horizontal scaling requires system concurrency because work must be handled across resources. It also fits a broader strategy of decoupling: complex problems can be divided into simpler, linked components. section 1.6 section 1.3
