The core of a cloud provider’s offerings is the ability to provision instances. An instance is similar in concept to a virtual machine, which is an emulation of a particular computer system. While historically virtual machines were thought of as running on a specific piece of hardware—perhaps a server in a company’s machine room—a cloud instance is thought of as running somewhere unspecified within the cloud provider’s vast infrastructure. Precisely where and how an instance is provisioned is often not revealed to you as a customer, although you do have some coarse-grained control (see “Regions and Availability Zones” for one example). All that matters to you is that you ask for an instance, and the cloud provider brings it to life.
Instances running in Azure are called “virtual machines.”
The features of an instance, beyond its precise physical location, are up to you. You can choose from a variety of combinations of CPU power and memory, any of several operating systems, and different storage options, just to start with. Once the instance has been provisioned, you can access it over SSH through the networking capabilities of the cloud provider. From that point on, it acts and feels just like a real, physical machine. You have root access, so you can install applications, perform additional upgrades, start web servers or Hadoop daemons, and so on.
Your cloud provider account has limits that affect the number and size of instances you can provision. ...