Global and Local Load Balancing

Introduction

Load balancing operates at two distinct scales: global and local.

• Global server load balancing (GSLB): Distributes traffic across multiple geographic regions.

• Local load balancing: Distributes traffic within a specific data center to improve resource efficiency and server utilization.

The following sections detail these techniques.

Global server load balancing

GSLB selects a serving location using policies such as geography, latency, available capacity, and health. If one location fails, the policy may direct new traffic elsewhere. The transition depends on detection, routing, and any cached destination information.

Global traffic management can be implemented on-premises or through a managed service. Define the failure domain it handles; routing between regions and failing over between zones are related but distinct design choices.

In the illustrated design, local load balancers report health and capacity information to a global control plane. That plane uses the reports and regional policy to choose destinations. Other implementations obtain health information differently, so treat this connection pattern as one architecture.