Comprehensive Overview of Load Balancing Technologies and Algorithms

Load balancing is a network technique that distributes incoming requests across multiple servers in a cluster, presenting a single virtual IP (VIP) to clients while transparently forwarding traffic to the appropriate backend machine.

The article explains the four main layer‑based categories: Layer 2 (MAC‑based), Layer 3 (IP‑based), Layer 4 (transport‑layer TCP/UDP), and Layer 7 (application‑layer HTTP, DNS, etc.), describing how each determines the forwarding method.

Commonly used software load balancers are introduced: LVS (Linux Virtual Server) for high‑performance Layer 4 balancing, Nginx for flexible Layer 7 balancing, and HAProxy for reliable TCP/HTTP proxying and high availability.

The article then classifies load‑balancing algorithms into static (e.g., Round Robin, Random, Hash) and dynamic (e.g., Least Connections, Fastest Response, Observed, Predictive, Dynamic Ratio, QoS, ToS, Rule‑based). Each algorithm’s principle, advantages, disadvantages, and typical use cases such as read‑only database traffic, cache key consistency, or real‑time service quality are discussed.

Additional concepts include weighted distribution, priority groups, consistency hashing, range‑based and modulo‑based key distribution, as well as hybrid approaches that combine DNS, Layer 4, and Layer 7 techniques for large‑scale internet services.

Overall, the article serves as a practical guide for architects and engineers designing scalable, high‑availability infrastructures.