# Regional Load Balancer

> Regional load balancers on GCP charge for each forwarding rule by the hour regardless of traffic, so a rule left behind after an environment teardown bills for zero requests indefinitely. ZopNight discovers regional balancers via Cloud Asset Inventory and flags rules that served no backend traffic over its metrics lookback.

Source: https://zop.dev/integrations/gcp/gcp-load-balancer
Updated: 2026-08-19

---

A regional load balancer distributes traffic within a single region for internal or external workloads. Its forwarding rules carry hourly charges independent of traffic volume.

## Forwarding rules are the regional LB meter

A load balancer is not one billable object but a chain of them, and the part that meters is the forwarding rule. Each rule accrues an hourly charge from the moment it exists, entirely independent of whether a single request flows through it. Data processing adds a usage-based component on top when traffic does flow, but the floor cost (rules multiplied by hours) is fixed. A regional balancer serving heavy production traffic and one serving a decommissioned staging environment can carry the same rule charges; only one of them is buying anything.

## ZopNight's traffic-silence check for regional rules

ZopDev discovers regional load balancers via Cloud Asset Inventory and flags rules with no backend traffic over the metrics lookback. That silence test is the core signal: an hourly-billed rule whose backends received nothing across the observation window is either pre-production, misconfigured, or abandoned, and each of those deserves a human look. Load balancers have no stop operation, so nothing here is schedulable. The savings motion is identifying dead rules and deleting them.

## Regional balancers after environment teardowns

The characteristic waste pattern is the incomplete teardown. Deleting the VMs of a test environment does not delete the balancer in front of them; the forwarding rule, target proxy, and backend service all survive with an empty or unhealthy backend list, and the rule keeps its hourly meter running. Internal load balancers are especially prone to this because nothing external ever notices they stopped serving. A second pattern is duplication: two rules created during a migration where one was meant to be temporary, both still billing a year later.

## Auditing regional load balancing in the console

Google Cloud console → Network services → Load balancing lists every balancer with its scope and backend summary. Open a suspect balancer and check its monitoring tab: a flat zero request graph across 30 days, paired with an hourly-billed forwarding rule, is the cleanup signature worth acting on.