Load Balancing

MiniGateway supports multiple load balancing algorithms to distribute requests across upstream targets.

Available Algorithms

Algorithm	Description	Best For
round-robin	Distribute evenly in rotation	Equal-capacity backends
least-connections	Route to server with fewest active requests	Varying load patterns
hash	Consistent hashing based on request key	Session affinity
health-aware	Combine least-connections with health checks	Production deployments

Algorithm Details

Target Weight

Targets can have different weights for weighted distribution:

{
  "upstreamId": "upstream_123",
  "target": "http://server-a:5000",
  "weight": 3 // Receives 3x more requests
}

With weights [A:3, B:1] and round-robin:

• Requests: A→A→A→B→A→A→A→B…

Target Health

The load balancer tracks target health:

• Healthy - Successfully responding
• Unhealthy - Failed responses or timeouts

Unhealthy targets are automatically excluded from selection until they recover.

Creating Load Balancers

You can create load balancers programmatically:

import { createLoadBalancer } from "@minigateway/core";

const lb = createLoadBalancer("round-robin", {
  targets: [
    { url: "http://server-a:5000", weight: 1 },
    { url: "http://server-b:5000", weight: 2 },
  ],
});

const target = lb.select();

Configuration Example

Complete upstream with targets:

// Create upstream
{
  "name": "backend-pool",
  "algorithm": "health-aware"
}

// Add targets
{
  "upstreamId": "upstream_xyz",
  "target": "http://10.0.0.1:5000",
  "weight": 1
}
{
  "upstreamId": "upstream_xyz",
  "target": "http://10.0.0.2:5000",
  "weight": 2
}

Next Steps

• Upstreams Configuration - Configure upstreams and targets
• Proxy Engine - Request processing flow
• Plugins - Modify request routing behavior