Cloud concepts

What cloud computing is

Cloud computing is renting computing services (servers, storage, databases, networking, software, analytics) over the internet from a provider, and paying for what you use instead of buying and running your own hardware. The provider owns and maintains the datacenters. You get resources in minutes through a portal, a command line or an API.

Key characteristics worth knowing:

• On-demand self-service: you create resources yourself without raising a ticket.
• Broad network access: reachable over the internet or private links.
• Resource pooling: the provider shares hardware between many customers (multi-tenancy) with logical isolation.
• Rapid elasticity: capacity grows and shrinks quickly.
• Measured service: usage is metered, so you pay for what you consume.

Shared responsibility model

Security and operations are split between the provider and the customer. Microsoft always owns the physical layer: datacenter buildings, power, cooling, physical hosts and the physical network. The customer always owns their data, their accounts and identities, and the devices and endpoints that connect to the cloud. Everything in between shifts depending on the service type.

Rules of thumb for exam questions:

• The more managed the service, the more Microsoft takes on. The customer is never fully off the hook: data and identity stay with you in every model.
• In IaaS, you patch the guest operating system. In PaaS and SaaS, Microsoft does.
• Physical security of the datacenter is always Microsoft’s, in every cloud model except a private cloud you run yourself.

Cloud models

Related ideas:

• Multi-cloud means using more than one public cloud provider. It isn’t the same as hybrid.
• Azure Arc and Azure Stack are the Microsoft answers for hybrid: manage and run Azure-style services on your own hardware or other clouds.
• Cloud bursting is a hybrid scenario: run normally on-premises, then add public cloud capacity during a spike.

Choosing a model

Consumption-based model and pricing

In the consumption-based model you pay only for what you use, with no upfront hardware purchase and no cost when you stop a resource that doesn’t bill when idle.

Comparing pricing models:

Moving from CapEx to OpEx is a benefit for finance teams: lower upfront cost, costs that follow demand, and no paying for idle capacity. But a badly managed cloud can cost more than on-premises, which is why cost management matters (see the management and governance notes).

Serverless

Serverless doesn’t mean there are no servers. It means you don’t manage them. The provider handles provisioning, scaling and patching, and you pay only while your code runs.

• Typical Azure examples: Azure Functions (event-triggered code), Azure Logic Apps (workflows), and serverless tiers of databases.
• Characteristics: event-driven, scales automatically (including to zero), billed per execution or resource-second, no VM to maintain.
• It is a PaaS-style option sitting above containers and VMs in terms of how much you manage.
• Good for short, event-triggered tasks (resize an image when uploaded, process a queue message). Poor fit for constantly running, heavy, or long-lived processes where a VM or container is cheaper.

Benefits of cloud services

High availability and scalability

• High availability means the service stays up despite failures. Azure delivers it through redundancy: availability zones, region pairs, load balancing and service level agreements (SLAs), which state the uptime Microsoft commits to (for example 99.9% or 99.99%).
• Scalability means adjusting resources to demand.
  • Vertical scaling (scale up/down): give an existing machine more or less CPU and memory.
  • Horizontal scaling (scale out/in): add or remove instances.
  • Autoscale does this automatically from a metric or schedule.
• Elasticity is scalability that happens automatically and quickly in both directions, so you don’t over-provision.

Reliability and predictability

• Reliability is the ability to recover from failure and keep working. Cloud provides it through multiple regions, zones, backups and disaster recovery services.
• Predictability covers performance (autoscaling and load balancing keep response times steady) and cost (you can forecast and set budgets using tools such as the pricing calculator and cost analysis).

Security and governance

• The platform provides physical security, built-in threat protection and compliance certifications.
• Governance is the set of rules and controls that keep resources consistent and compliant: policies for allowed configurations, role-based access, tagging and auditing.
• The cloud makes it possible to apply those rules everywhere at once, and to meet standards such as ISO 27001 or regional regulations.

Manageability

Two kinds:

• Management of the cloud: the cloud resources themselves can be managed automatically, for example autoscaling, deploying from templates and monitoring that triggers alerts or fixes.
• Management in the cloud: your tools to manage resources, such as the portal, command line, PowerShell, REST APIs and automation.

Summary of benefits

Cloud service types

Ways to tell them apart in a question:

• “Install and patch the OS yourself” or “need administrator access to the server”: IaaS.
• “Developers deploy code and the platform handles scaling and patching”: PaaS.
• “Users just sign in and use it”: SaaS.
• Serverless is a flavour of PaaS where you don’t even size the platform.

Traps

• Azure Virtual Machines is IaaS even though Microsoft runs the host. You still manage the guest OS.
• Azure SQL Database is PaaS; SQL Server on a VM is IaaS. Same engine, different responsibility.
• A hybrid cloud is not the same as having both IaaS and PaaS. Hybrid is about mixing on-premises and public environments.
• Reserved pricing is not consumption-based in the pure sense: you commit to a term. Pay-as-you-go is the purest consumption model.