Prepare and test your skills

Prepare and test your skills

Physical versus software-defined security is the difference between protecting a fixed building and protecting every user and device with software rules. Traditional on-premises security relies on physical perimeters and hardware appliances at the network edge. Google Cloud (opens in a new tab) instead uses software-defined, identity-based controls that work across its global infrastructure.
Physical perimeter security in traditional data centers uses layered physical safeguards and hardware appliances to guard one location. Physical safeguards include electronic access cards, alarms, vehicle barriers, perimeter fencing, metal detectors, biometrics, laser-beam intrusion detection, 24/7 cameras, and security guards. Access to the data center floor is further restricted by a security corridor that uses multi-factor access control with badges and biometrics. Hardware appliances at the network edge inspect and filter traffic, which creates a trust boundary where users and devices inside the perimeter may be treated as more trusted than those outside.
Software-defined and identity-based security in Google Cloud protects access without depending on a single network perimeter. The zero-trust model (opens in a new tab) removes perimeter-based trust, so no user or device is inherently trustworthy and every access request must be authenticated and authorized regardless of origin. Access decisions can consider user role, device posture, and location, and they grant least-privilege access only to necessary resources. The model also uses an assume-breach mindset, which assumes an attacker may already be inside and therefore requires continuous verification and monitoring. Identity-Aware Proxy (IAP) and Chrome Enterprise Premium enforce access based on user identity and context, moving security from the network edge to individual users and devices.
Google runs its own services such as Search, YouTube, and Gmail on the same infrastructure that customers use. Customers therefore benefit from the same global-scale security controls and practices that protect Google services. This shared platform means protection travels with the identity and the workload, not just with the building.
The shared responsibility model (opens in a new tab) defines who secures what in Google Cloud and changes where trust boundaries sit. Google is responsible for protecting the cloud itself, including the underlying infrastructure, while the customer is responsible for securing what they bring to the cloud, including data. In infrastructure as a service (IaaS), Google is responsible only for hardware, storage, and network, while in software as a service (SaaS), Google is responsible for everything except data and its access and usage. Customers no longer manage physical data center security or hardware appliances for the provider infrastructure. Instead, they focus on software-defined controls such as identity, access policies, network segmentation, and logging.
Google Cloud controls that replace hardware appliances are software-defined network and security services that work in a distributed, cloud-first way. Other hybrid and multicloud security considerations (opens in a new tab) describe how these controls connect on-premises, Google Cloud, and multicloud environments. The controls are organized into peer categories:
Comparing physical and software-defined approaches shows differences in trust, controls, scope, and responsibility. Well-Architected Framework security guidance (opens in a new tab) frames these differences as a move from building-centered protection to policy-centered protection. The table compares the main dimensions:
| Dimension | Traditional on-premises security | Google Cloud security |
|---|---|---|
| Primary trust boundary | Physical perimeter and network edge | Identity and context of each access request |
| Main control style | Physical safeguards and hardware appliances | Software-defined policies and identity-based controls |
| Scope | Data center and its network perimeter | Global network and distributed cloud features |
Exam tip: In a zero-trust model, being inside the network does not make a user or device trusted; every access request must be authenticated and authorized.
Operational visibility and automation is the shift from manual, physical security operations to continuous, automated oversight in the cloud. In a traditional data center, teams patch hardware by hand, perform physical walk-through audits, and review scattered logs after an incident. In Google Cloud, Security Command Center (opens in a new tab) provides centralized visibility where automated detectors check configurations all the time, findings flow to one place, and compliance status updates daily.
Security Health Analytics is the detection service in Security Command Center that automatically checks cloud resources for misconfigurations and vulnerabilities. Each check is performed by a detector, and each detector maps to controls in standards such as Payment Card Industry Data Security Standard (PCI DSS) v3.2.1 and National Institute of Standards and Technology (NIST) SP 800-53. Because checks run continuously across projects, folders, and organizations, teams see drift right away instead of waiting for the next manual review. Predefined posture templates bundle the detectors for a specific standard and can be deployed without changes.
Example detectors from the PCI DSS and NIST templates show what continuous monitoring watches:
PUBLIC_DATASET checks whether a dataset is open to public access.AUDIT_LOGGING_DISABLED checks whether audit logging is turned off for a resource.OPEN_SSH_PORT checks whether a firewall has an open SSH port that allows generic access.MFA_NOT_ENFORCED checks whether a user is not using 2-step verification.KMS_KEY_NOT_ROTATED checks whether rotation for Cloud Key Management Service encryption is not turned on.SSL_NOT_ENFORCED checks whether a Cloud SQL database instance does not use SSL for all incoming connections.When a detector finds a problem, data flows from the detector to a finding record in Security Command Center. The workflow depends on the detector staying enabled and the resource staying in scope, so turning off a detector or moving a resource removes that visibility.
A finding in Security Command Center is a record of a vulnerability or misconfiguration that names the affected resource and explains how to fix it. Findings focus on observable controls under customer or shared responsibility, such as identity, firewall, encryption, and logging settings, rather than the physical infrastructure that Google manages. This boundary reflects shared responsibility, where Google secures the data centers and the customer secures what is configured inside them. Each finding carries context needed to prioritize and fix the issue, including a description, a recommendation, the numerical control ID, and remediation steps.
Remediation follows an ordered flow from detection to improved compliance:
A three-step process flow showing how a Security Health Analytics detector reports a misconfiguration as an active finding in Security Command Center, the team remediates the resource, and the compliance score updates, with active findings lowering the score and muted findings excluded.
Only after the configuration is corrected does the compliance percentage improve.
Centralized logging and audit visibility in Google Cloud means log generation, protection, and network visibility are checked as configuration, not by visiting machines. Detectors confirm that audit logging is enabled for resources, that VPC Flow Logs are turned on for network visibility, and that a locked retention policy is set so logs cannot be changed or deleted early. If logging is disabled or retention is not locked, Security Command Center raises a finding before the gap causes a blind spot during investigation. Logs from many projects flow to centralized storage where the retention policy applies to the stored logs, and Essential Contacts configuration ensures notifications reach the right people. The result is continuous evidence that logging exists and is protected, rather than point-in-time proof from a manual audit.
Evaluating compliance across the environment means viewing the percentage of supported controls that pass for a standard at a chosen scope. The Compliance page in the Google Cloud console and the Risk Overview page in the Security Operations console show this status for the organization, folder, or project. Compliance reports are generated daily, so leaders track trends over time instead of relying on a single audit snapshot. Security Command Center reports only on the subset of controls that can be observed automatically, because some controls cannot be automated.
Where Security Command Center is activated controls what compliance view is available:
| Activation scope | What compliance view shows | How to see more detail |
|---|---|---|
| Project level | Compliance statistics only for the activated project | Switching to a parent folder or organization does not display the page |
| Organization level | Compliance statistics for the entire organization, including folders and projects | Switch to a specific folder or project to see its statistics |
The predefined posture template for PCI DSS (opens in a new tab) and the vulnerability findings index (opens in a new tab) document which detectors and findings support these compliance views.
Exam tip: Fixing all Security Command Center findings does not guarantee full regulatory certification, because compliance reporting monitors configuration and does not replace a formal compliance audit.
The shared responsibility model is a security framework in which the cloud provider secures the underlying infrastructure and the customer secures data, configurations, and access. In Google Cloud, Google manages physical data centers, hosts, and default encryption, while the customer manages identity, permissions, and service settings. In traditional on-premises environments, the customer owns every layer from hardware to data.
Provider responsibilities for cloud infrastructure cover security of the cloud itself, including physical facilities, hosts, and baseline protections. For Managed Service for Apache Airflow (Managed Airflow), provider duties include encrypting data at rest by default with Google-owned and Google-managed keys and running environment cluster nodes on Shielded VMs, which are hardened against rootkits and bootkits. The provider also supplies basic security features such as uniform bucket-level access and private IP networking options. Underlying resources are configured to follow VPC Service Controls (VPC SC) rules when those perimeters are used. These duties show that Google handles the hardware and platform foundations that customers never touch directly.
Customer responsibilities for data and configuration cover security in the cloud, meaning everything the customer places or sets in Google Cloud. For Managed Airflow, customer duties include assigning Identity and Access Management (IAM) roles and permissions and setting Airflow UI Role-Based Access Control (RBAC) levels for different users. Customers apply Domain Restricted Sharing (DRS) policies to limit access to selected domains and configure Customer Managed Encryption Keys (CMEK) (opens in a new tab) when they want more control over encryption keys. They also configure Private IP networking and VPC SC perimeters, enable audit logging, SSL enforcement, key rotation, multi-factor authentication, and firewall rules, and deploy the PCI DSS posture template in Security Command Center to check Payment Card Industry Data Security Standard compliance. Each task secures the service configuration and data, not the underlying hardware.
Comparing provider and customer duties makes the split concrete by showing who manages each security area. Assured Workloads framework guidance (opens in a new tab) maps similar controls to regulatory expectations, while cloud threat detection guidance (opens in a new tab) shows how threats against customer-configured resources are observed. The table uses Managed Airflow and related Google Cloud services as examples:
| Security area | Provider manages | Customer manages |
|---|---|---|
| Encryption at rest | Default encryption with Google-managed keys | Optional CMEK for more key control |
| Compute hosts | Shielded VMs for cluster nodes | No direct host management |
| Network access | Private IP networking capability | Configuring VPC network and VPC SC perimeters |
| Identity and access | IAM service availability | Assigning IAM roles, Airflow RBAC, and DRS policies |
| Compliance monitoring | Security Command Center platform | Deploying the PCI DSS posture template |
The dependency is clear: the customer configuration depends on the provider platform, but the provider cannot secure customer choices such as who gets access or which keys are used.
Why the model matters for cloud security is that it moves the customer focus from physical perimeters to identities, configurations, and data. In on-premises environments, the customer must secure everything, including hardware and network appliances. In the cloud, the provider secures the infrastructure, so the customer can focus on access policies, encryption choices, and service settings. This division reduces operational burden but does not remove responsibility for secure configuration. A misconfigured firewall, identity role, or logging setting remains the customer duty even though Google protects the data center around it.
Exam tip: Default encryption uses Google-managed keys, but choosing CMEK moves key control to the customer.
Traditional on-premises security relies on physical safeguards and hardware appliances at the network edge to guard one location, where users and devices inside the perimeter may be treated as more trusted. Google Cloud security uses software-defined, identity-based controls that authenticate and authorize every access request regardless of origin, so protection travels with the identity and the workload across global infrastructure.
The zero-trust model removes perimeter-based trust so no user or device is inherently trustworthy and being inside the network does not make a user or device trusted. Every access request must be authenticated and authorized based on identity and context such as user role, device posture, and location, granting least-privilege access only to necessary resources with continuous verification and monitoring.
Google is responsible for protecting the cloud itself including the underlying infrastructure, physical facilities, hosts, and default encryption, while the customer is responsible for securing what they bring to the cloud including data, configurations, identity, and access. In infrastructure as a service (IaaS) Google is responsible only for hardware, storage, and network, while in software as a service (SaaS) Google is responsible for everything except data and its access and usage.
Security Health Analytics uses detectors to continuously check cloud resources for misconfigurations and vulnerabilities mapped to controls in standards such as Payment Card Industry Data Security Standard (PCI DSS) v3.2.1 and National Institute of Standards and Technology (NIST) SP 800-53. When a detector finds a problem it creates a finding record that names the affected resource and explains how to fix it, and active findings lower the compliance percentage until remediated while compliance reports are generated daily.
A financial services company is migrating its infrastructure from an on-premises data center to Google Cloud. Currently, their security operations team relies on the following traditional practices:
The company wants to modernize its security operations by adopting cloud-native operational visibility and automation.
How should the company transform its security operations in Google Cloud?
Last updated