Public, Private, and Hybrid Cloud

Cloud deployment models define where the infrastructure for your applications and data resides, who manages it, and how you access it. The three primary models are public, private, and hybrid. Each model addresses different business needs regarding control, security, scalability, and cost.

In a public cloud, a third-party provider like Microsoft Azure owns and manages the entire infrastructure—servers, storage, networking, and data centers. As a customer, you provision and use resources (virtual machines, databases, etc.) over the internet, paying only for what you consume (pay-as-you-go pricing).

How it works: - Azure operates massive data centers around the world. These data centers house thousands of servers, storage arrays, and network equipment. - You create an Azure subscription, and within that subscription, you can create resources like VMs, web apps, or databases using the Azure portal, CLI, or APIs. - Azure abstracts the underlying hardware. You do not know which physical server your VM runs on, nor do you care. Azure manages hardware failures, maintenance, and upgrades. - You can scale resources up or down instantly. For example, during a holiday sales spike, you can add more VM instances to handle traffic, then remove them when traffic subsides.

Key characteristics: - Multi-tenant: Multiple customers share the same physical infrastructure, but their data and workloads are isolated using virtualization and network segmentation. - Elasticity: Resources can be scaled automatically based on demand. - No capital expenditure: You pay operational expenses (OpEx) only for what you use. - High availability: Azure offers SLAs (e.g., 99.9% uptime for VMs) and provides built-in redundancy across regions.

Business problem solved: A startup with limited capital needs to launch a web app quickly without buying servers. Public cloud allows them to spin up infrastructure in minutes and pay as they grow.

A private cloud is a cloud environment dedicated to a single organization. It can be hosted on-premises (in your own data center) or by a third-party provider exclusively for you. The key is that the infrastructure is not shared with other organizations.

How it works: - You own or lease the hardware (servers, storage, network). You install a hypervisor (like Hyper-V or VMware) to virtualize resources. - You deploy a cloud management platform (like Azure Stack HCI or VMware vSphere) to provide self-service provisioning, orchestration, and metering. - Users within your organization can request VMs or storage through a portal, and the platform automatically provisions them on the available hardware. - You are responsible for all maintenance: patching OS, replacing failed disks, upgrading firmware, managing power and cooling.

Key characteristics: - Single-tenant: No resource sharing with other organizations. - Full control: You decide hardware, software, security policies, and compliance measures. - Higher cost: You bear capital expenditure (CapEx) for hardware and ongoing operational costs for staff, power, and cooling. - Limited scalability: Scaling requires purchasing and installing new hardware, which can take weeks.

Business problem solved: A bank must keep customer financial data on-premises due to regulatory compliance. A private cloud allows them to get some benefits of cloud (self-service, automation) while maintaining physical control.

A hybrid cloud combines public and private clouds, allowing data and applications to be shared between them. The two environments are connected via a secure network (typically VPN or Azure ExpressRoute) and operate as a unified infrastructure.

How it works: - You run some workloads on-premises (private cloud) and others in Azure (public cloud). - You establish connectivity: either a site-to-site VPN over the internet or a dedicated private connection via Azure ExpressRoute. - You can use Azure services to manage and monitor both environments from a single pane of glass (e.g., Azure Arc, Azure Monitor). - Workloads can be moved between environments based on need. For example, you might keep sensitive data on-premises but use Azure for compute-intensive analytics on that data.

Key characteristics: - Flexibility: You choose where to run each workload based on cost, compliance, latency, or other factors. - Consistent platform: Azure provides tools (like Azure Stack HCI and Azure Arc) that make the on-premises environment feel like an extension of Azure. - Complexity: You must manage two environments, network connectivity, and data synchronization. - Cost optimization: You can use public cloud for burst capacity instead of over-provisioning your private cloud.

Business problem solved: A retailer runs its core transaction system on-premises for low latency but uses Azure for seasonal peak traffic and analytics. Hybrid cloud allows seamless scaling without abandoning existing investments.

| Feature | Public Cloud | Private Cloud | Hybrid Cloud | |---------|--------------|---------------|--------------| | Ownership | Cloud provider | Organization | Both | | Scalability | Instant, elastic | Limited by hardware | Elastic via public cloud | | Cost Model | OpEx only | CapEx + OpEx | Mixed | | Control | Low | High | Medium | | Security | Shared responsibility | Organization's sole responsibility | Shared with segregation | | Example Use Case | Dev/test, web apps | Regulated data, legacy apps | Bursting, disaster recovery |

Azure supports all three models: - Public: Standard Azure services (VMs, App Service, SQL Database). - Private: Azure Stack HCI (hyperconverged infrastructure for on-premises) or Azure Private Edge Zones. - Hybrid: Azure Arc (management across environments), Azure ExpressRoute (dedicated private connection), and Azure Backup / Azure Site Recovery (disaster recovery to Azure).

The exam will expect you to know the definitions, advantages, and disadvantages of each model. You should also understand that hybrid cloud is not a separate type of cloud but a combination of public and private.

Scenario 1: E-Commerce Startup Using Public Cloud A new online retailer launches a platform for handmade goods. They have no existing IT infrastructure and a limited budget. They choose Azure public cloud. They deploy a web app using Azure App Service, store product images in Azure Blob Storage, and use Azure SQL Database for inventory. During Black Friday, traffic spikes 10x. Azure Autoscale automatically adds more App Service instances, and the database scales up. After the sale, resources scale down, and they pay only for what they used. The startup avoids buying servers that would sit idle most of the year. If they had chosen private cloud, they would have needed to purchase enough capacity for peak traffic, wasting capital.

Scenario 2: Hospital Using Private Cloud for Patient Records A hospital must store electronic health records (EHR) on-premises due to government regulations that data cannot leave the country. They build a private cloud using Azure Stack HCI in their data center. Doctors and nurses access the system via a local portal. The hospital retains full control over data and compliance. However, they also run a public-facing appointment booking system in Azure public cloud to handle variable patient demand. The two systems are not connected, so no hybrid model is needed. If they had used public cloud for EHR, they would violate compliance.

Scenario 3: Manufacturer Using Hybrid Cloud for Burst Computing A manufacturing company runs its ERP system on-premises in a private cloud. Twice a year, they run complex supply chain simulations that require massive compute power for a week. Instead of buying extra servers for these peaks, they use Azure public cloud. They set up a site-to-site VPN between their on-premises network and Azure. When a simulation starts, they deploy hundreds of VMs in Azure, run the simulation, then tear them down. The ERP data stays on-premises, but the compute is done in the cloud. This hybrid approach saves 60% compared to provisioning for peak. Common mistakes: forgetting to secure the VPN connection or not monitoring data transfer costs, which can be significant.