By Cloudledger · Updated
One resource, several billing meters
People compare a Pricing Calculator compute rate with the total charge for a virtual machine and conclude the bill is wrong. Usually it is not: a running VM is billed through several separate meters. Compute hours, operating-system licensing, each attached managed disk, snapshots, public IP addresses, outbound data transfer, backup storage and monitoring can all appear as distinct lines. Read the bill at meter level before deciding anything looks incorrect.
Group the cost by resource and meter
In Cost Management, filter to the VM’s resource group or resource ID and group by meter. You will normally see compute separated from storage and networking. Disks usually bill by provisioned capacity and performance tier rather than by how full they are, so an almost-empty 1 TB disk can cost as much as a full one.
- Filter costs to the resource or resource group for a complete period.
- Group by meter category and meter to see every component.
- Note quantity and unit for each meter before comparing rates.
- Check whether the disks, IPs and backups belong only to this VM.
Licensing is often the missing line
Windows Server and some Linux distributions include licence charges in addition to compute. If your organisation has eligible licences, Azure Hybrid Benefit may apply, but eligibility, coverage and terms come from your licensing agreement rather than from the portal view alone. Confirm entitlement with whoever owns your licensing before assuming a discount applies.
Stopped is not always free
A VM stopped from inside the guest operating system can remain allocated and continue to incur compute charges. Deallocating the VM in Azure stops normal VM compute billing, but attached disks, reserved IP addresses, snapshots and backups continue to be billed. Check the power state in Azure rather than relying on what the guest reports.
A worked read of one VM
Illustrative example: a VM shows ₹10,000 of compute, ₹2,500 of managed disks, ₹800 of backup and ₹400 of bandwidth in a month. The total of ₹13,700 is roughly 37% above the compute figure alone. Nothing is wrong with the bill; the estimate simply covered one meter. These figures are for illustration and are not Azure price quotes.
Turn the finding into an action
Once you know which meter dominates, the action differs. High compute suggests rightsizing or scheduling; high disk cost suggests reviewing capacity, tier and orphaned disks; high bandwidth suggests examining traffic paths. Cloudledger’s Cost Explorer can show these meters together for a single resource, and Compute Intelligence can inform a sizing decision.