Study guide · Domain 3.0 · 18% of the exam
Security Architecture: the SY0-701 objectives in plain English
Domain 3, Security Architecture, makes up 18% of the SY0-701 exam, so roughly one question in six comes from this material. It covers how systems are built and connected, how data is protected wherever it sits, and how an organization keeps running when something fails. More than any other domain, it rewards understanding who is responsible for what: you, your organization, a cloud provider, or a vendor.
What the objectives cover
3.1 Compare and contrast security implications of different architecture models.
This objective compares where systems run and what each choice means for security. On-premises infrastructure gives you full control and full responsibility. Cloud services split the work under the shared responsibility model: the provider secures the platform, while you keep your data, accounts, and configuration, and the split shifts across IaaS, PaaS, and SaaS. Hybrid setups mix both, and third-party vendors add risk you manage rather than control.
You compare virtualization with hypervisors and containers, Internet of Things devices, industrial control systems and SCADA, embedded systems, infrastructure as code, microservices, and serverless designs. Centralized designs are simpler to run, decentralized designs limit blast radius, and air gaps physically isolate sensitive systems. Every option trades availability, cost, scalability, and risk transference.
3.2 Given a scenario, apply security principles to secure enterprise infrastructure.
Given a scenario, this objective asks where each control goes and why. Segmentation divides the network into zones and VLANs so one compromise cannot spread, and placement decides which traffic each control sees.
Firewalls filter traffic by rules and access control lists. A next-generation firewall inspects applications and users, a web application firewall shields web apps, a unified threat management box combines several functions, and a stateful firewall tracks connection state. An IDS watches and alerts, while an IPS sits in line and blocks. A VPN encrypts traffic over untrusted networks for remote users and site to site links. You meet 802.1X and EAP access control, jump servers, port security, load balancers, sensors, and SASE using TLS or IPsec.
3.3 Compare and contrast concepts and strategies to protect data.
Know what you hold. Data types include regulated data, trade secrets, intellectual property, legal records, and financial data, each with its own rules. Labels such as sensitive, confidential, public, restricted, private, and critical set how strongly you protect data and who may see it. Data sits in three states, at rest on storage, in transit across a network, and in use during processing, and each state needs protection.
The methods are encryption, which makes data unreadable without a key, hashing for integrity fingerprints, masking to hide part of a value, tokenization to swap real data for a token, and obfuscation to make data hard to read. Geographic restrictions, segmentation, and permission levels control where data lives and who reaches it.
3.4 Explain the importance of resilience and recovery in security architecture.
Surviving failure is the point. Load balancing and clustering provide high availability, spreading work so nodes take over for each other. Recovery sites trade cost for speed: hot sites run at once, warm sites need setup, cold sites start bare, and geographic dispersion guards against disasters.
Backups are a last line of defense: copies onsite for restores and offsite so one event cannot destroy both, frequency set by tolerable data loss, snapshots for rollback, and encryption on the files. Replication copies data continuously elsewhere. Tabletop exercises, failover tests, simulations, and parallel processing prove the plan, and recovery testing confirms backups restore. Generators and a UPS cover power, platform diversity limits flaws, and continuity of operations keeps the business running.
Worked example
A company migrates its email system to a SaaS platform. Under the shared responsibility model, which of the following tasks is the responsibility of the cloud provider rather than the customer?
With SaaS, the provider runs and maintains the application itself, including the servers it runs on and the patching of that software. The customer keeps responsibility for its own users, data, and configuration choices.
Why the other options are wrong
- AManaging user accounts and performing access reviews: Account management and access reviews always stay with the customer, because only the customer knows who should have access.
- CConfiguring mailbox retention and archiving policies: Retention and archiving settings are customer configuration decisions made inside the service.
- DTraining employees to recognize phishing messages: Security awareness training for employees is an organizational responsibility that no cloud model transfers to the provider.
Acronyms in this domain
| Acronym | What it means |
|---|---|
| IaC | Infrastructure as Code |
| IoT | Internet of Things |
| ICS | Industrial Control System |
| SCADA | Supervisory Control and Data Acquisition |
| NGFW | Next-Generation Firewall |
| WAF | Web Application Firewall |
| IDS | Intrusion Detection System |
| IPS | Intrusion Prevention System |
| VPN | Virtual Private Network |
| VLAN | Virtual Local Area Network |
| SASE | Secure Access Service Edge |
| UPS | Uninterruptible Power Supply |
| RTO | Recovery Time Objective |
| RPO | Recovery Point Objective |
Where the help desk meets this domain
You already work inside this domain every day, even if your tickets never use its vocabulary. When your company moves email to a SaaS platform and you spend the next month creating accounts, resetting passwords, and removing access for people who left, that is 3.1 shared responsibility in action: the provider runs the service, but account management and access reviews stay with you. When a user asks why the VPN will not connect from home, or why a branch office cannot reach a new application, you are troubleshooting 3.2, because VPNs, firewall rules, and segmentation decide what traffic is allowed to flow. When you connect a new printer, camera, or badge reader to the IoT segment instead of the workstation network, you are applying the same objective, and when a new hire cannot join the corporate Wi-Fi because the device is not enrolled, that is 802.1X doing its job. When you reach a server through a jump server instead of logging in directly, you are using the controlled path that 3.2 describes. None of that daily work moves to the vendor, no matter what the migration announcement promised.
The rest of the domain shows up in quieter ways. When you encrypt a laptop before handing it out, you are protecting data at rest under 3.3. When a user asks whether a spreadsheet can be emailed to a personal account, the answer comes from its classification, also 3.3, and when you mask a card number while reading it back to a caller, you are using one of that objective's methods by hand. When you tell a user to send a file through the approved sharing link instead of a personal email attachment, you are protecting data in transit and following its classification at the same time. When you restore a deleted file from last night's backup, or confirm that a restore actually worked before you close the ticket, that is 3.4, and it is why backup frequency and recovery testing matter more than a green check mark on the backup job. When the ticketing system fails over during an outage and nobody notices, that is high availability from 3.4 working as designed. You are also the person users call when a failover, a certificate warning, or a blocked website interrupts their morning, which is why this domain reads less like theory than the others. Help desk work is where architecture decisions meet real users, so the scenarios in this domain should feel familiar.
Traps to watch for
The biggest trap in this domain is assuming a cloud move transfers work that it never transfers. In every model, including SaaS, account management, access reviews, and configuration choices stay with the customer. The provider patches its own platform, but it does not decide who works at your company, who should keep access after changing roles, or how you set retention and sharing policies. Any answer that hands user accounts to the provider is wrong for the same reason every time. A related trap reverses the models: with IaaS the provider supplies the infrastructure, but you still patch the guest operating system and every application you install on it, so questions that give the provider that job in an IaaS scenario are testing whether you know where the line sits in each model. If a question stem says the provider will handle something, check which model is named before you believe it. Insurance and contracts can transfer some financial risk under 3.1, but they never transfer the work of securing your own accounts and data.
When a device cannot protect itself, look for the compensating control. A camera, thermostat, or industrial controller that cannot run an agent and rarely takes patches is not made safe by trying harder on the device itself. The exam answer is to isolate it, usually on its own segment with tightly restricted access, and the same logic covers embedded systems and SCADA equipment that cannot be taken offline for updates. Finally, keep the recovery site order straight: hot, warm, and cold sites trade cost against recovery speed in that order, so the cheapest option is always the slowest to bring back. Watch for answers that call a warm site instant, treat a snapshot as a complete backup, or describe hashing as a way to read data back later, because each one swaps a partial measure for the real thing. One more distinction the exam loves: an IDS only detects and alerts, while an IPS blocks, so a scenario that needs traffic stopped in line is never answered with an IDS alone.
Now try it. The 5 free sample questions for Domain 3 are in practice mode, with full explanations.
Practice Domain 3 questions