Mastering Load Balancing: Principles, Types, and Algorithms Explained

When facing massive user traffic, high concurrency, and large data volumes, high‑performance servers, large databases, and efficient programming languages (e.g., Go, Scala) are essential. Once a single machine reaches its limits, business splitting and distributed deployment become necessary to handle large‑scale website traffic.

1. Load Balancing Principles

System scaling can be vertical (adding CPU, memory, disk) or horizontal (adding more machines). Horizontal scaling forms a cluster with a load balancer that distributes requests across multiple servers, improving performance, fault tolerance, and scalability.

Application Cluster: Deploy the same application on multiple machines to form a processing cluster that receives requests from the load balancer.

Load Balancer Device: Distributes user requests to one of the servers in the cluster based on a load‑balancing algorithm.

Load balancing solves several problems:

Alleviates concurrency pressure and increases throughput.

Provides fault‑tolerance for high availability.

Enables horizontal scalability by adding or removing servers.

Offers security features such as filtering and black‑/white‑listing.

2. Load Balancing Classification

Based on implementation technology, load balancing can be divided into DNS, HTTP, IP, layer‑2, and hybrid types.

2.1 DNS Load Balancing

Uses multiple A records in DNS to map a domain to several servers, forming a cluster. It is often used as the first level of load balancing for large sites.

Advantages:

Simple to use; offloads balancing work to DNS servers.

Improves performance by directing users to the nearest server.

Disadvantages:

Poor availability due to DNS propagation delays.

Limited scalability; control resides with the domain registrar.

Limited visibility into server health and fewer algorithms.

Practice: Use DNS as the first tier, pointing A records to an internal load balancer that then distributes traffic to real web servers. Suitable for internet companies but not for complex business systems.

2.2 IP Load Balancing

Operates at the network layer by modifying the destination IP address of incoming packets. The load balancer selects a real server IP, rewrites the packet’s destination, and forwards it. Responses are similarly rewritten before reaching the client.

Advantages: Performs distribution in the kernel, offering better performance than application‑level solutions.

Disadvantages: All traffic passes through the load balancer, making its network interface a potential bottleneck.

2.3 Layer‑2 Load Balancing

Modifies MAC addresses at the data‑link layer without changing IP addresses. Real servers share a virtual IP with the load balancer, allowing direct client responses and avoiding the load balancer’s bandwidth limit (direct routing mode).

Advantages: High performance.

Disadvantages: Complex configuration.

Practice: Direct routing (DR) mode is widely used for its performance benefits.

2.4 Hybrid Load Balancing

Combines different load‑balancing methods across multiple server groups to achieve optimal performance. Two typical patterns are illustrated:

Pattern 1: Separate static and dynamic content; a reverse‑proxy cluster caches static resources while dynamic requests are forwarded to an application load balancer.

Pattern 2: Handles dynamic requests with a dedicated load‑balancing layer.

3. Load Balancing Algorithms

Common algorithms include:

3.1 Round‑Robin

Distributes requests sequentially across servers. Works well when servers have identical hardware.

3.2 Random

Assigns requests randomly; simple but unsuitable for heterogeneous servers.

3.3 Least Connections

Sends traffic to the server with the fewest active connections, adapting to current load.

3.4 Hash (Source‑IP)

Hashes the client IP to consistently route a user’s session to the same server, enabling session stickiness.

3.5 Weighted

Extends other algorithms by assigning weights to servers, allowing finer control over traffic distribution.

4. Hardware Load Balancing

Dedicated hardware appliances (e.g., F5, A10) provide high‑performance load balancing but are costly. When evaluating hardware solutions, consider:

Comprehensive feature support across layers and global balancing.

Performance capable of handling tens of thousands of concurrent connections.

Stability proven by large‑scale deployments.

Integrated security functions such as firewalls and DDoS protection.

Ease of management, support, and maintenance.

Cost, which can range from hundreds of thousands to over a million RMB for premium models.

Drawbacks: High price and limited scalability compared to software solutions.

Most internet companies prefer open‑source software load balancers (e.g., Nginx) for cost‑effectiveness, reserving hardware balancers for critical, high‑traffic scenarios.