Microsoft Azure Fundamentals Training Course

Cloud computing is the delivery of computing services — servers, storage, databases, networking, software, and analytics — over the internet, on demand, and paid for by what you use. That single sentence is the foundation of everything else on AZ-900, and the exam will test whether you can apply it rather than recite it. So it is worth slowing down on what each part of it actually means.

Rented, not owned

The traditional way to run an application is to buy the hardware. You size a server for the busiest hour you expect in the next three years, purchase it, wait for it to arrive, rack it in a room you also pay for, and then own it whether it is busy or idle. The cloud replaces that purchase with a rental. Microsoft owns the datacenter, the racks, the power, the cooling, the network, and the physical servers. You rent a slice of that capacity for as long as you need it and hand it back when you do not.

The consequence is that capacity becomes a decision you can revisit rather than a commitment you are stuck with. If you provision a virtual machine and discover it is too small, you resize it. If a workload is only needed at month end, you run it for two days and stop paying for it the other twenty-eight. Nothing about that is possible when the asset sits in a room you own.

On demand and self-service

The second characteristic is that you get resources without asking a human. You sign in to the Azure portal, or issue a command, and a virtual machine exists a few minutes later. There is no purchase order, no procurement cycle, no delivery date. This is usually described as self-service, on-demand provisioning, and it is the property that makes the cloud fast rather than merely cheap.

It also means that provisioning becomes something anyone with permission can do, which is why governance — the subject of the third module of this course — exists at all. When the friction of buying a server disappears, the cost control has to be moved somewhere else.

A shared pool of resources

Behind a cloud service is a very large pool of physical hardware shared across many customers. Your virtual machine runs on a physical host that is also running other customers' virtual machines, isolated from each other by the hypervisor. Pooling is what makes the economics work: Microsoft can run its datacenters at high utilization because your quiet hours are somebody else's peak. You get access to a scale of infrastructure no single organization would build for itself, and Microsoft gets a utilization rate no single organization would achieve.

This is also the honest answer to the phrase "the cloud is just someone else's computer." It is accurate but incomplete. The cloud is someone else's computers, at a scale, redundancy, and geographic spread that changes what you can build on them.

Measured and metered

Cloud services are metered. Compute is typically billed per second or per hour that a resource is running; storage by the gigabyte stored per month plus the number of read and write operations; networking largely by the volume of data leaving the platform. The meter is the mechanism that turns a capital purchase into an operating expense, and it is the reason cost management is a first-class discipline in Azure rather than an afterthought.

Metering has a behavioral consequence the exam likes: because you pay for what is running, turning things off saves money, and leaving things running wastes it. A virtual machine that is deallocated stops accruing compute charges. A disk attached to it does not — storage is still storage whether the machine is on or off.

What Azure is

Azure is Microsoft's cloud platform: a set of more than two hundred services running in Microsoft-operated datacenters around the world. You reach them through the Azure portal in a browser, through command line tools, through code, or through automated deployment files. Some of those services give you a machine to manage yourself; others give you a running platform where you supply only your application; others give you finished software you simply use. Telling those apart is the subject of the next several lessons.

What the exam is testing here

AZ-900 is a vocabulary and discrimination exam. For this topic that means you should be able to look at a described situation and say which characteristic of cloud computing it illustrates. A retailer that handles Black Friday traffic without buying servers is illustrating elasticity and the consumption model. A startup that ships a product in six weeks because it never waited on hardware is illustrating self-service provisioning and agility. A company that keeps a copy of its data in a second geography is illustrating the geographic reach of a global platform.

Two cautions. First, do not confuse cloud computing with virtualization. Virtualization is a technology the cloud uses heavily, but a virtualized server in your own basement is not cloud computing, because it is neither rented, nor self-service, nor metered. Second, do not treat "the cloud" as a single thing you are either on or off. Most real organizations run some workloads in Azure, some on-premises, and some in another provider, and the exam expects you to be comfortable with that mixture rather than surprised by it.