Loading...
The traffic director that spreads requests across servers so none of them drowns.
Quick Definition
A load balancer distributes incoming traffic across multiple servers or services to improve reliability, performance, and availability. In the cloud, load balancers are managed services priced by usage and data processed, making them both an availability tool and a cost factor.
A load balancer is a component that distributes incoming requests across multiple servers. It is what lets a service run as many identical replicas: users hit one address, the balancer spreads them out, performs health checks, and quietly stops sending traffic to instances that fail.
Load balancers make both high availability and autoscaling possible, since neither works without something to spread traffic across a changing set of machines. In the cloud they are managed services billed hourly plus per traffic processed, which creates a classic small-leak pattern: balancers from deleted projects keep billing, and architectures that give every tiny service its own balancer multiply a modest fee into a real line item.
Example. An audit finds 19 load balancers with zero healthy targets, relics of finished projects, costing about $400 monthly for routing nothing. A monthly orphan check becomes part of the team's cleanup routine.
On Kubernetes, an ingress controller lets one balancer serve many services. The Cloud Cost Management Guide covers hunting these small recurring leaks.
Layer 4 routes by network connection, fast and simple. Layer 7 understands HTTP and can route by path, host, or headers.
Individually no, collectively yes. Hourly fees plus traffic charges across dozens of balancers add up, and orphaned ones are pure waste.
Periodic health checks. Instances that fail checks are removed from rotation until they recover.