# Azure API Management

> Azure API Management bills hourly by tier and unit count, and Premium units run into thousands of dollars monthly whether the gateway serves 1 request or millions. ZopNight inventories each instance's tier and units via Resource Graph, reads 60 days of request metrics, and surfaces tier-downgrade opportunities on quiet gateways.

Source: https://zop.dev/integrations/azure/apimanagement
Updated: 2026-08-19

---

API Management publishes, secures, and meters APIs through a managed gateway, billed hourly by tier; Premium units run into thousands of dollars monthly. Non-production APIM instances on production tiers are frequent oversizing victims.

## An hourly gateway meter that ignores traffic

API Management bills by tier and unit count for every hour the instance exists, and the meter is indifferent to request volume: a Developer-tier gateway handling steady traffic and a Premium unit serving nothing each bill exactly what their tiers dictate. Premium units run into thousands of dollars monthly, which makes the tier choice the single largest cost decision for this resource type, and one that is frequently made once at setup and never revisited.

## Tier and unit inventory ZopNight builds

Discovered via Azure Resource Graph with tier and unit counts. Azure Monitor metrics over a 60-day lookback show request volume, Cost Management billing attributes spend, and tier-downgrade opportunities surface in review. APIM instances are not schedulable, because a gateway cannot be stopped and restarted on a timer without breaking every API behind it. ZopNight's leverage is visibility: putting what each instance costs next to how many requests it actually serves.

## Production tiers parked in dev subscriptions

The recurring waste shapes: a Premium instance created for a proof of concept and never downgraded, billing production rates for a handful of test calls; capacity units added during a traffic spike and never removed after it passed; and dev or staging instances mirroring the production tier when only production needs its throughput and SLA. Because APIM instances take a long time to provision, teams keep them around just in case, which is precisely how a heavy hourly meter runs unattended for months.

## Checking an APIM instance's tier in the portal

Azure portal → API Management services lists every instance with its tier. Open an instance and the Locations blade shows its unit count, while the request metrics on the same instance answer whether that capacity has been earning its hourly rate.