Load Balancer

Azure Load Balancer provides a layer 4 load balancer with support for both TCP and UDP protocols. It can be deployed for both internal and public (external) clients, and a single LB can contain any number of routing rules.


  • Frontend IP configurations allow mapping public and private IP addresses to the load balancer.
  • Backend pools specify targets (NICs or IP address-VNet pairs) to which traffic can be sent.
  • Health probes define rules for testing the health of a backend, including a check frequency and failure count before a node is considered unhealthy.
  • Load balancing rules define how to direct traffic from a given IP configuration and port combination to a backend pool.
  • Inbound NAT rules allow directing traffic to a specific backend node directly, useful for troubleshooting.


Instance limit1001,000
Backend pool members?Single availability set or scale setAny blend of VMs, availability sets, or scale sets
Zone redundant frontend?
Multiple frontend IP configurations?
NSG required?
Provisioning time30-90 seconds<30 seconds

Traffic distribution modes

Traffic distribution can be bound to either:

  • Hash-based affinity, where the load balancer will use a 5-tuple hash, best for availability:
    • Source IP
    • Source port
    • Destination IP
    • Destination port
    • Protocol
  • Session affinity, best for e.g. RD Gateway:
    • Source IP
    • Destination IP

Cross region load balancer

Cross-region load balancer is designed to provide high availability across regions without having to use DNS round robin services such as Traffic Manager. It's currently in preview and not generally available.