Architecture Models: Cloud, Virtualization, IoT & ICS
Security+ SY0-701 · Security Architecture
Comparing architecture models
Objective 3.1 asks you to compare the security implications of different ways of building IT systems. The exam does not expect you to be a cloud architect, but it does expect you to know who is responsible for security in each model, where the risks are, and which model fits a given requirement. The key idea running through all of them is the shared responsibility model: in every model except a fully on-premises one, security duties are split between the organization and a provider, and the split moves depending on the model.
Cloud service models: IaaS, PaaS, SaaS
The three cloud service models differ in how much the provider manages and how much you manage. In Infrastructure as a Service (IaaS), the provider supplies virtual machines, storage, and networking, and you manage the operating system, applications, and data — you keep the most control and the most security responsibility. In Platform as a Service (PaaS), the provider also manages the operating system and runtime, and you manage only your applications and data. In Software as a Service (SaaS), the provider manages nearly everything and you use the application through a browser; your security responsibility is mostly limited to access control, data classification, and configuration. As you move from IaaS to SaaS, the provider takes on more of the security work, but you also lose visibility and control. A useful memory aid: the more "as a service" you buy, the less you manage — and the more you must trust the provider's security.
Cloud deployment models describe who the cloud belongs to. A public cloud is shared infrastructure owned by a provider and available to anyone. A private cloud is dedicated to one organization, hosted on premises or by a provider. A community cloud is shared by several organizations with common requirements (such as government agencies). A hybrid cloud combines public and private clouds, letting an organization keep sensitive workloads private while using public cloud for the rest. Multicloud means using more than one public cloud provider, which avoids vendor lock-in but increases management complexity.
Virtualization, containers, and serverless
Virtualization runs multiple virtual machines on one physical host using a hypervisor. Type 1 hypervisors run directly on the hardware (such as VMware ESXi or Microsoft Hyper-V); Type 2 hypervisors run on top of an operating system (such as VirtualBox). The security risks include VM escape, VM sprawl, and the fact that the hypervisor itself becomes a single point of failure — if it is compromised, every virtual machine on the host is at risk. Containers share the host operating system's kernel rather than running a full guest operating system, which makes them lighter and faster but means a kernel vulnerability can affect all containers on the host. Serverless computing runs code in response to events without the customer managing any servers at all; the provider handles the infrastructure, and the customer's responsibility shrinks to the code and its permissions — but misconfigured permissions and insecure code remain the customer's problem.
IoT and industrial control systems
The Internet of Things (IoT) covers the huge range of connected devices that are not traditional computers: cameras, sensors, smart locks, medical devices, and industrial equipment. IoT devices are a major security headache: they often ship with default credentials, cannot be patched, have limited processing power for encryption, and are deployed in large numbers that nobody tracks. The exam expects you to know that IoT devices should be isolated on their own network segment, have default credentials changed, and be inventoried like any other asset.
Industrial control systems (ICS) run the physical world: power plants, water treatment, factories, and building management. The most important ICS type is SCADA (supervisory control and data acquisition), which monitors and controls industrial processes over large areas. ICS environments have special constraints: availability is the top priority (a power plant cannot go down for patching), systems may be decades old and unpatchable, and the protocols they use (such as Modbus) were designed without security. The exam tests the principle that ICS networks should be isolated from the corporate network and the internet — often with an air gap or a data diode — and that patching an ICS system requires careful testing because an unplanned restart can cause physical harm.
Infrastructure as code and other models
Infrastructure as code (IaC) defines infrastructure in version-controlled files rather than by manual configuration. The security benefits are significant: configurations are consistent, changes are reviewed and auditable, and insecure settings can be caught by automated scanning before deployment. The risk is that a single bad template can be deployed everywhere at once, so templates must be reviewed and tested like any other code. On-premises architecture keeps everything in the organization's own data center: maximum control, but also full responsibility for physical security, patching, and availability. Managed service providers (MSPs) and managed security service providers (MSSPs) take on some or all of an organization's IT or security operations, which shifts responsibility but never eliminates the organization's accountability for its data.
Choosing the right model for a requirement
The exam often gives a business requirement and asks which architecture model fits best. Work through the trade-offs: if the requirement is maximum control over security and data, on-premises or a private cloud is the answer; if it is lowest management overhead, SaaS wins; if it is rapid scaling for unpredictable demand, public cloud or serverless fits; if it is keeping sensitive data in-house while using cloud for the rest, hybrid cloud is the answer. If the requirement mentions avoiding vendor lock-in, multicloud is the model. If it mentions consistent, auditable configuration, infrastructure as code is the answer. And if the scenario involves devices with limited security capabilities — cameras, sensors, medical equipment — the answer is almost always to isolate them on their own segment, because you cannot fix the device itself. The key skill is matching the requirement to the model's strengths rather than memorizing a ranking of models.
One more comparison the exam tests: centralized vs decentralized architecture. A centralized model (one data center, one identity provider, one management console) is easier to secure consistently but creates a single point of failure and a single high-value target. A decentralized model spreads risk but is harder to manage and monitor. Most real organizations are hybrids of both, and the exam's scenarios usually reward the answer that balances control with resilience rather than the extreme in either direction.
Exam traps to watch for
- Mixing up the cloud service models. In IaaS you manage the OS; in PaaS the provider manages the OS and you manage the app; in SaaS you manage almost nothing except access and data.
- Confusing containers with virtual machines. Containers share the host kernel; VMs run a full guest OS. A kernel flaw affects all containers on a host.
- Forgetting that ICS prioritizes availability over everything else. You cannot just reboot a power plant to install a patch.
- Thinking the cloud removes your security responsibility. In every cloud model, the customer always owns its data, its access control, and its configuration.
Practice questions
Test yourself on this guide. Answers are checked instantly, with a full explanation for each question.
Q1.A company moves its email to a cloud service where the provider manages the servers, the operating system, and the email application, and the company's administrators only manage user accounts and mail settings. Which cloud service model is this?
Explanation: In Software as a Service (SaaS), the provider manages the infrastructure, operating system, and application, and the customer only manages its own use of the application, such as accounts and settings. Option C is wrong because in IaaS the customer manages the operating system and applications. Option D is wrong because in PaaS the customer manages its own applications on the provider's platform. Option A is wrong because FaaS is a serverless model for running individual functions, not a complete email application.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models
Q2.A development team wants to deploy its own applications without managing operating systems or patching servers, but it needs control over the application code and its data. Which cloud service model best fits?
Explanation: Platform as a Service (PaaS) gives developers a managed platform where the provider handles the operating system and runtime, while the team controls its own applications and data. Option C is wrong because SaaS provides a finished application, not a platform for deploying custom code. Option B is wrong because IaaS requires the team to manage operating systems and patching. Option D is wrong because private cloud is a deployment model, not a service model, and does not by itself remove OS management.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models
Q3.A hospital deploys hundreds of network-connected patient monitors and infusion pumps. The devices have limited processing power, cannot run antivirus software, and the vendor rarely releases firmware updates. What is the most important security measure for these devices?
Explanation: IoT and medical devices with limited capabilities and rare updates should be isolated on their own network segment so that a compromised device cannot reach other systems and cannot be reached from the general network. Option D is wrong because the devices lack the processing power to run EDR agents. Option C is wrong because devices like infusion pumps do not support MFA logins. Option B is wrong because most IoT devices have no hard drive to encrypt, and disk encryption would not address the network exposure.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models
Q4.A water treatment plant uses a SCADA system to monitor pumps and valves. The system runs on a 15-year-old server that cannot be patched without shutting down water treatment for the region. Which security approach is most appropriate?
Explanation: ICS and SCADA systems prioritize availability and often cannot be patched, so the standard approach is to isolate the control network from the corporate network and the internet and strictly control who can access it. Option A is wrong because exposing a SCADA system to the internet is extremely dangerous. Option B is wrong because antivirus alone does not protect an unpatched, network-exposed control system. Option D is wrong because shutting down water treatment for a region is an unacceptable operational impact; isolation is the practical control.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models
Q5.A company defines all of its server configurations in version-controlled template files, and every change is reviewed and tested before it is deployed automatically. What is this practice called?
Explanation: Infrastructure as code (IaC) defines infrastructure in version-controlled files, making configurations consistent, reviewable, and auditable. Option A is wrong because SaaS is a cloud service model for finished applications. Option C is wrong because microsegmentation divides a network into isolated zones. Option D is wrong because container orchestration manages the deployment and scaling of containers, not infrastructure configuration in general.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models
Q6.A company runs its public website on a cloud provider's shared infrastructure and keeps its customer database in its own on-premises data center, with a secure connection between the two. Which deployment model is this?
Explanation: A hybrid cloud combines public cloud services with private (on-premises) infrastructure, connected so workloads can span both. Option A is wrong because the database is not in the public cloud. Option B is wrong because the website is not in a private cloud. Option D is wrong because a community cloud is shared by several organizations with common requirements, which is not the case here.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models
Q7.A security team is evaluating whether to run an application in containers or in virtual machines. Which statement about containers is correct?
Explanation: Containers share the host operating system's kernel, which makes them lightweight and fast to start, but means a kernel vulnerability can affect all containers on the host. Option D is wrong because running a full guest OS describes virtual machines, not containers. Option A is wrong because virtual machines provide stronger isolation than containers. Option B is wrong because containers do not use a hypervisor at all.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models
Q8.In a cloud environment using the shared responsibility model, which security responsibility ALWAYS remains with the customer, regardless of whether the service is IaaS, PaaS, or SaaS?
Explanation: In every cloud model, the customer always owns its data, its classification of that data, and who is allowed to access it. Option C is wrong because the provider patches the hypervisor in all cloud models. Option D is wrong because physical data center security is always the provider's responsibility. Option B is wrong because the provider maintains its own network hardware.
Exam objective: SY0-701 Objective 3.1: Compare and contrast security implications of different architecture models