Is an app with minimum replicas above zero billed when it is idle?

Yes. On the Consumption plan, replicas at the minimum count are billed for their vCPU and memory, at a reduced idle rate only while they handle no requests and use almost no CPU or network. Only scaling to zero stops these charges.

How many Container Apps environments should each client have?

Microsoft suggests one environment for related services that share a network, logs or Dapr configuration, and separate environments where apps need isolation, such as test and production. Each environment using Dedicated profiles carries its own fixed management charge.

Do ingress and networking choices affect the bill?

Yes. Request charges apply to HTTP requests arriving from outside the environment, though health probes are not billed and each subscription gets 2 million requests free each month. Private endpoints can add the Dedicated plan management charge.

Coverage

Azure Container Apps cost optimisation for MSPs

Microsoft's guidance for Container Apps names two common sources of waste: apps left running when they don't need to be, and CPU and memory allocations bigger than workloads need. The fix is to rightsize regularly and let intermittent apps scale to zero, so you pay only while they are in use.

9

automated finders for Container Apps

Read-only, across every connected client tenant, tied to real billing data.

Modernise
Remove unused
Rightsizing
Scaling

Where Container Apps spend goes to waste

The main cost levers Microsoft documents for Container Apps, and the ones an MSP should check across every client.

Match each workload to the right plan

The Consumption plan scales on demand, can scale to zero and bills only while apps run. The Dedicated plan bills per workload profile instance rather than per app, and can work out cheaper for steady, higher-scale deployments. ^[2]

Let intermittent apps scale to zero

Container Apps defaults to a minimum of zero replicas, and an app scaled to zero incurs no usage charges. Audit any app whose minimum has been raised without a clear reason, and keep a non-zero minimum only where an instance must always be running. ^[3]

Rightsize CPU and memory per replica

Consumption charges are metered per second on the vCPU and memory allocated to each running replica, so generous allocations raise the bill even when the code barely uses them. Compare allocations with real utilisation and trim them back. ^[4]

Deactivate revisions you don't need

In multiple revision mode, several revisions can stay active at once, each running its own replicas. Inactive revisions are not charged, so deactivate old ones once traffic has moved, or use single revision mode. ^[5]

Run finite tasks as jobs

Scheduled, batch and event-triggered work is better run as a Container Apps job than as an always-on app. On the Consumption plan, jobs are billed at the active rate only until each execution completes, with no idle charges. ^[4]

Consolidate apps onto shared profiles

Dedicated profiles are billed by the number of profile instances running, and several apps can share one instance. Putting compatible apps on shared profiles with sensible minimum and maximum counts avoids paying for half-empty nodes. ^[6]

How Cerberus Cloud covers Container Apps

Cerberus Cloud runs 9 automated finders for Container Apps across every connected client tenant. Each one ties a technical signal to real billing data, so findings arrive with an estimated recoverable value, ranked by impact, effort and risk, with evidence and remediation scripts for your engineers to review.

Access is read-only, so nothing changes in a client environment until your team decides it should. See how it works or the multi-tenant portfolio view.

Modernise
Remove unused
Rightsizing
Scaling

Frequently asked questions

Is an app with minimum replicas above zero billed when it is idle?: Yes. On the Consumption plan, replicas at the minimum count are billed for their vCPU and memory, at a reduced idle rate only while they handle no requests and use almost no CPU or network. Only scaling to zero stops these charges.
How many Container Apps environments should each client have?: Microsoft suggests one environment for related services that share a network, logs or Dapr configuration, and separate environments where apps need isolation, such as test and production. Each environment using Dedicated profiles carries its own fixed management charge.
Do ingress and networking choices affect the bill?: Yes. Request charges apply to HTTP requests arriving from outside the environment, though health probes are not billed and each subscription gets 2 million requests free each month. Private endpoints can add the Dedicated plan management charge.

See Container Apps findings on a sample portfolio

A 15-minute walkthrough: tenant connection, prioritised findings, remediation scripts and savings history.

Book a Walkthrough All coverage

Sources

Last checked 15 October 2026. Product details change; tell us if something here is out of date.

All 15 coverage families