Skip to main content

Load Balancing

GoZen supports several provider selection strategies beyond basic failover. You can choose a strategy per profile, combine it with health checks, and steer traffic based on availability, latency, or cost.

Available strategies​

Failover​

Try providers in order until one succeeds. This is the default strategy and works well for primary/backup setups.

{
  "profiles": {
    "default": {
      "providers": ["primary", "backup"],
      "strategy": "failover"
    }
  }
}

Round robin​

Distribute requests evenly across multiple equivalent providers.

{
  "profiles": {
    "balanced": {
      "providers": ["provider-a", "provider-b", "provider-c"],
      "strategy": "round-robin"
    }
  }
}

Least latency​

Prefer the provider with the lowest recent response time.

{
  "profiles": {
    "fast": {
      "providers": ["us-east", "us-west", "eu"],
      "strategy": "least-latency"
    }
  }
}

Least cost​

Prefer the cheapest provider for the requested model.

{
  "profiles": {
    "budget": {
      "providers": ["cheap-provider", "premium-provider"],
      "strategy": "least-cost"
    }
  }
}

Health-aware routing​

All strategies can work with health monitoring. When health_aware is enabled, unhealthy providers are skipped automatically until they recover.

{
  "profiles": {
    "production": {
      "providers": ["primary", "secondary", "tertiary"],
      "strategy": "least-latency",
      "health_aware": true
    }
  }
}

Choosing a strategy​

  • Use failover for reliability-first routing.
  • Use round-robin when providers are interchangeable.
  • Use least-latency for interactive or time-sensitive workloads.
  • Use least-cost when budget matters more than raw speed.
  • Profiles explains how provider groups are defined.
  • Routing covers scenario-based provider selection.
  • Health Monitoring explains how health checks affect routing.