For a better understanding of how Azure VMs work, you need to know the statuses of the VMs.
State |
Description |
Running | The VM is running and you get charged for usage as usual |
Stopped | The VM is shut down by Windows/Linux, but you still get charged for the VM, as it is still deployed to the same physical host and resources are still reserved for it |
Stopped (Deallocated) | The VM is stopped by the stop button on the VM plane via the Azure portal; this state will cost the user no charges, except for the storage of the VM |
At the moment there are two billing models available: Pay-As-You-Go and Reserved Instances.
- Pay-As-You-Go: This model enables you to pay for the VMs' usage as long as you are using them. In this model, Microsoft charges by the minute, so if you have used a VM for 61 minutes, for example, you will only pay for those 61 minutes.
- Reserved Instances: This model enables you to reserve a VM for 1, 2, or 3 years, which offers a high cost reduction compared to the Pay-As-You-Go model. The cost savings can be up to 82%.
For more information about Reserved Instances, you can check out the following page: https://azure.microsoft.com/en-us/pricing/reserved-vm-instances/.
At the time of writing, Microsoft has two Service Level Agreements (SLAs) for Azure VMs, they are as follows:
- Two or more VMs within the same availability set have 99.95% availability to one VM guaranteed
- Using a single VM that uses Premium Storage will provide at least 99.9% availability
To stay updated on Microsoft SLAs for Azure VMs, check the following page: https://azure.microsoft.com/en-us/support/legal/sla/virtual-machines/v1_6/.