AWS ALB LCU / NLB NLCU Calculator

LCU and NLCU cost is a max-driver problem, not a sum-of-traffic problem

This page exists to answer one question: which hourly driver actually dominates the load balancer bill? New connections, active connections, processed bytes, and rule evaluations do not all stack equally here. The billing logic is driven by the highest usage dimension.

• Connection drivers: new and active connection behavior that often dominates chatty or bursty traffic.
• Processed bytes: the payload-size dimension that can overtake everything during large-transfer periods.
• Rule evaluations: the ALB-specific driver that can matter more than traffic volume when rule count and request rate rise together.

Where LCU and NLCU estimates usually drift

• Teams average all drivers together even though one hourly maximum determines the bill.
• Processed bytes are modeled, but bursty connection behavior or rule evaluations are ignored.
• One calm hour is used as the baseline even though peak traffic shifts the dominant driver.
• Users assume more requests always means more cost, while the actual dominant dimension may be bytes or rule evaluations.

What to review before trusting the capacity-unit baseline

• Look at each driver separately and identify the max instead of blending them into one traffic story.
• Check peak hours and burst windows because that is where the dominant driver often flips.
• Model ALB rule evaluation pressure explicitly if routing complexity is part of the architecture.
• Use this page to estimate units per hour, then move to the main load balancer cost page for full pricing.

Baseline vs dominant-driver shift scenarios

How to review the first real capacity-unit month

• Check which driver actually dominated in production before changing all of the assumptions at once.
• Review hourly peaks, not just monthly averages, because max-driver billing hides inside those windows.

Next steps