The cost of a load balancer explained

Quick answer

A cloud load balancer bills two ways at once: a small hourly charge for the load balancer itself (billed whether or not it sees traffic) plus a usage charge based on capacity units that combine new connections, active connections, bandwidth, and rule evaluations. So cost scales with both uptime and traffic. A single always-on load balancer costs roughly a low double-digit dollar amount per month at minimum, before traffic. To keep cost low, consolidate services behind one load balancer with host or path routing rather than running many, and delete idle ones.

A load balancer feels like a small piece of infrastructure, but its cost model catches people out because it bills two ways simultaneously: a fixed hourly rate just for existing, plus a usage-based charge that scales with traffic. Understanding both halves is how you avoid paying for idle load balancers and how you keep the usage charge in check.

The two-part bill

Component	What it reflects
Hourly charge	The load balancer existing, billed even at zero traffic
Capacity units	New connections, active connections, bandwidth, rule evaluations

On AWS, Application and Network Load Balancers bill an hourly rate plus Load Balancer Capacity Units (LCUs), where an LCU is the highest of several dimensions (new connections per second, active connections, processed bytes, and rule evaluations). GCP and Azure use similar two-part models: a forwarding-rule or hourly base charge plus data processing. The key point: even an idle load balancer costs the hourly rate, so an unused one is pure waste.

Roughly what it costs

The hourly charge alone puts a single always-on load balancer at roughly a low double-digit dollar amount per month before any traffic. The capacity or data-processing charge then scales with load. For a busy service, the usage portion can exceed the base. Exact figures vary by provider and region (the same regional variation that shows up in data transfer out), so treat these as approximate and price your specific configuration.

Keeping load balancer cost low

The biggest saving is consolidation: instead of one load balancer per service, put many services behind a single Application Load Balancer using host-based or path-based routing rules. One load balancer with smart routing replaces several, cutting the fixed hourly charges. Delete idle load balancers (a common orphaned resource, they keep billing the hourly rate with nothing behind them). Put a CDN in front of cacheable traffic so fewer requests reach the load balancer at all, which lowers both its capacity units and your origin load.

Load balancer cost in context

For a small app, the fixed hourly charge dominates and consolidation is the main lever. For a high-traffic service, the usage charge dominates and reducing processed bytes (via caching and efficient responses) matters more. Either way, the load balancer is rarely the largest line, but idle ones are pure waste and a proliferation of per-service load balancers adds up. Model your load balancers against the resource catalog so their fixed and usage cost is visible before deploy, and fold it into your cost per customer math for services that run many of them.

FAQ

How much does a cloud load balancer cost?

It bills two ways: a fixed hourly charge for the load balancer existing (billed even at zero traffic), which alone runs roughly a low double-digit dollar amount per month, plus a usage charge based on capacity units that scale with new connections, active connections, bandwidth, and rule evaluations. For busy services the usage portion can exceed the base. Exact rates vary by provider and region.

What are Load Balancer Capacity Units (LCUs)?

On AWS, an LCU is the billing unit for the usage portion of Application and Network Load Balancers. It measures the highest of several dimensions: new connections per second, active connections, processed bytes, and rule evaluations. You pay for the number of LCUs consumed, so whichever dimension is highest for your traffic drives the usage cost, on top of the fixed hourly charge.

Does an idle load balancer still cost money?

Yes. The hourly charge bills whether or not the load balancer sees any traffic, so an idle load balancer with nothing behind it is pure waste, roughly a low double-digit dollar amount per month for nothing. Idle load balancers are a common orphaned resource; auditing for and deleting them is a quick cost win.

How do I reduce load balancer cost?

Consolidate: put many services behind a single Application Load Balancer using host-based or path-based routing instead of one load balancer per service, cutting fixed hourly charges. Delete idle load balancers. Put a CDN in front of cacheable traffic so fewer requests reach the load balancer, lowering its capacity units. For most small apps, consolidation is the biggest lever.

Can one load balancer serve multiple services?

Yes. An Application Load Balancer supports host-based and path-based routing rules, so a single load balancer can route to many backend services based on the hostname or URL path. This replaces several per-service load balancers with one, cutting the fixed hourly charges. The main constraint is rule limits and the operational simplicity of separate load balancers, which some teams prefer despite the cost.

Does C3X price load balancers before deploy?

Yes. C3X reads your Terraform and prices resources against a live catalog, so load balancers and their fixed hourly cost are visible before deploy. This helps catch a proliferation of per-service load balancers or idle ones in the pull request, where consolidating them is cheap, rather than discovering the accumulated hourly charges on the bill.

What to do next

See what your load balancers cost before they ship. C3X reads your Terraform and prices your resources against a live catalog. Start with the quickstart.