Load balancing

Load balancing distributes incoming traffic across multiple backend servers, improving availability and scalability. Quark supports several load balancing algorithms and can be configured to balance traffic for any service.

Basic configuration

To set up load balancing, you need to:

1. Define a load balancer with a list of backend servers
2. Reference that load balancer in a location’s target

[loadbalancers.my_backends]
backends = ["192.168.1.10", "192.168.1.20", "192.168.1.30"]

[services.my_app]
domain = "example.com"

[[services.my_app.locations]]
source = "/*"
target = "http://${my_backends}:8080"

In this example, requests to example.com will be distributed across the three backend servers on port 8080.

Load balancing algorithms

Quark supports different algorithms for distributing traffic:

Round Robin (default)

Distributes requests evenly across all backends in sequential order:

[loadbalancers.my_backends]
algo = "round_robin" # Optional since it's the default
backends = ["192.168.1.10", "192.168.1.20", "192.168.1.30"]

Each backend receives an equal share of requests in rotation.

Weighted Round Robin

Allows you to assign different weights to backends, sending more traffic to servers with higher capacity:

[loadbalancers.my_backends]
algo = "round_robin"
backends = ["192.168.1.10", "192.168.1.20", "192.168.1.30"]
weights = [5, 3, 2]

In this configuration:

• 192.168.1.10 receives 50% of requests (5/10)
• 192.168.1.20 receives 30% of requests (3/10)
• 192.168.1.30 receives 20% of requests (2/10)

IP Hash

Routes requests from the same client IP to the same backend server, providing session persistence:

[loadbalancers.my_backends]
algo = "ip_hash"
backends = ["192.168.1.10", "192.168.1.20", "192.168.1.30"]