Load balancing algorithms. How to choose a specific service instance?

Load balancing is the distribution of incoming requests across multiple service instances to improve fault tolerance and performance.

Main load balancing algorithms:

• Round Robin — requests are sequentially directed to each instance in a circular manner.
• Least Connections — the instance with the fewest active connections is chosen.
• IP Hash — the instance is selected based on a hash of the client's IP, ensuring session persistence on the same server.
• Weighted Round Robin / Least Connections — consider the weight of the instance (e.g., capacity).

The choice of a specific instance depends on the requirements:

• If uniform load distribution is important — Round Robin.
• If instances differ in capacity — Weighted algorithms.
• To maintain session persistence — IP Hash.
• To minimize delays — Least Connections.

Example in C# (simplified Round Robin):

class LoadBalancer
{
    private readonly List<string> _instances;
    private int _lastIndex = -1;

    public LoadBalancer(List<string> instances)
    {
        _instances = instances;
    }

    public string GetNextInstance()
    {
        _lastIndex = (_lastIndex + 1) % _instances.Count;
        return _instances[_lastIndex];
    }
}

This class cyclically returns the next instance from the list.