The company is deploying AI applications on a Google Kubernetes Engine (GKE) cluster, characterized by complex, microservices-based architecture. The primary concerns are protecting against data poisoning and lateral movement between containerized components.

Prisma AIRS Network Intercept (C) directly addresses the concern of lateral movement between containerized components. Prisma Cloud’s runtime network security capabilities monitor and control network traffic between containers and pods, detecting and preventing unauthorized communication, suspicious network patterns, and other activities indicative of lateral movement within the GKE cluster. By limiting an attacker’s ability to move laterally, it significantly reduces the attack surface and helps contain breaches. While data poisoning can occur through various means, including APIs, preventing lateral movement is a foundational security measure that hinders attackers from reaching and manipulating data sources, thus indirectly contributing to data integrity.

Why other options are incorrect:

• A: AI Access Security with Advanced URL Filtering and B: AI Access Security with App-ID Cloud Engine: ‘AI Access Security’ is not a recognized Palo Alto Networks product. Advanced URL Filtering primarily protects against web-based external threats, and App-ID Cloud Engine identifies applications. Neither offers the comprehensive runtime protection specifically for internal lateral movement or direct data poisoning within a GKE microservices environment.
• D: Prisma AIRS API Intercept: While crucial for preventing data poisoning that occurs via malicious API calls (which is a significant vector for AI applications), API Intercept focuses specifically on the API layer. The question also explicitly mentions ‘lateral movement between containerized components,’ which is a broader network-level runtime threat more directly addressed by Network Intercept. Network Intercept provides a more foundational layer of defense against unauthorized movement within the cluster, which often precedes other attack objectives like data poisoning.

The core requirements for end-user path visibility, endpoint health metrics, centralized Zoom call quality analytics, and insights into supporting protocols like DNS are all directly addressed by Autonomous Digital Experience Management (ADEM).

ADEM agents deployed on end-user devices (Windows desktops in this case) provide deep visibility into the user’s experience, covering the endpoint itself, the network path, and application performance. ADEM has specific integrations for UCaaS applications like Zoom to collect call quality metrics. The GlobalProtect agent is commonly used to deploy and manage ADEM on endpoints, providing a unified platform for security and experience monitoring. Prisma Access serves as the SASE platform where GlobalProtect and ADEM data are often aggregated and managed, even for users in an office if their traffic is routed through it or for consistent management.

Option A is incorrect because while ADEM is relevant, an ION device is primarily for SD-WAN branch deployments, and the question emphasizes monitoring end-user experience on Windows desktops rather than branch network infrastructure. The critical component is ADEM on the endpoint.

Option C is incorrect because Prisma SD-WAN focuses on optimizing WAN link performance and application routing across the WAN. While it has an application dashboard and link quality monitoring, it does not provide the granular end-user path visibility, endpoint health metrics, or specific Zoom call quality analytics from the desktop perspective that ADEM offers.

Option D is incorrect because Prisma Browser or its extension with RUM metrics is limited to monitoring web application performance within the browser context. It would not provide insights into desktop applications like Zoom, overall endpoint health, or the broader network path for all applications.

✅ Option C (Correct)

The recommended approach for safely performing a critical software upgrade with minimal application downtime in a highly available cloud deployment is a “blue/green” deployment strategy. This involves provisioning a completely new, parallel Azure VMSS with the new PAN-OS version. Once this new fleet of firewalls is fully operational and validated, the Azure Load Balancer is updated to redirect traffic from the old VMSS to the new one. This ensures near-zero downtime as the old environment remains available until the new one is fully functional, and provides a quick rollback option by simply redirecting traffic back if issues arise.

❌ Why the other choices are incorrect:

• Option A is incorrect: Updating the image in an Azure VMSS and initiating an in-place upgrade of instances often involves taking instances offline or rebooting them, which can lead to downtime. It lacks the near-zero downtime and safety net of a blue/green deployment.
• Option B is incorrect: Configuring Azure Load Balancer probes is a fundamental aspect of high availability, ensuring traffic only reaches healthy instances. However, it is a prerequisite for HA, not the strategy for performing the upgrade itself with minimal downtime. The blue/green approach leverages these probes as part of the cutover process.
• Option D is incorrect: Azure Update Manager is primarily designed for managing updates to Azure VMs and guest operating systems, not typically for pushing vendor-specific appliance software like PAN-OS. More importantly, pushing updates simultaneously to all instances would inevitably cause significant downtime, directly contradicting the requirement for minimal application downtime.

The core requirement is to ensure consistent user-to-group mapping from Okta is available to Prisma Access for policy enforcement, regardless of the user’s location (home or branch office). Panorama, as the central management platform for Prisma Access, integrates with the Cloud Identity Engine (CIE) specifically for this purpose.

The Cloud Identity Engine is designed to connect with cloud identity providers like Okta, pulling user and group information and making it available to Palo Alto Networks security services, including Prisma Access. This allows for the creation of consistent, identity-aware security policies based on department-specific group memberships.

• Option A is incorrect: While a User-ID agent can collect user information, the Cloud Identity Engine is the modern, scalable, and recommended architecture for integrating cloud IdPs like Okta directly with Prisma Access for user and group mapping.
• Option C is incorrect: SAML federation is primarily for user authentication (SSO) and verifying identity during a request. It does not provide the persistent user-to-group mapping necessary for comprehensive and consistent policy enforcement across all traffic in the way CIE does. Policies require the group information to be known and applied consistently, not just for a single authentication event.
• Option D is incorrect: Configuring individual SD-WAN devices and GlobalProtect clients to directly query Okta for user information is not scalable, practical, or secure for this use case. Prisma Access needs this information centrally for policy evaluation, not clients or edge devices.

Option B provides the optimal solution by leveraging the core components of a full SASE architecture as described. Prisma Access remote networks allow global remote offices to connect securely and efficiently to the nearest Prisma Access cloud location via IPSec. From there, Prisma Access service connections offer direct, high-performance, and secure tunnels to the public cloud environment. This architecture inherently provides lowest latency due to Prisma Access’s global PoP network, highest throughput, and greatest resilience through redundant infrastructure. Crucially, it maintains a balance of threat inspection as all traffic passes through Prisma Access’s security services.

A is incorrect because Prisma Access Agent or PAC file is primarily for individual end-user devices, not the site-to-site connectivity required for remote offices with IONs. C is incorrect as deploying IONs within the cloud primarily extends SD-WAN into the cloud, but the primary secure and optimized path from global remote offices for SASE is via Prisma Access. D is incorrect because a direct private link from a private data center to the public cloud, while performant, can introduce latency for global remote offices and may bypass the comprehensive SASE security provided by Prisma Access for traffic originating directly from remote offices.

To strictly enforce that access requests to highly sensitive SaaS applications originate specifically from Prisma Browser and prevent standard web browsers, a unique identifier must be leveraged within the Microsoft Entra ID Conditional Access policy.

Option B (Unique device token or Device-ID issued by Prisma Browser and validated by Entra ID) is the most effective method. Secure enterprise browsers like Prisma Browser are designed to integrate with Identity Providers (IdPs) like Entra ID. This integration typically involves the browser registering as a client application and providing specific, verifiable identifiers or claims (such as an application ID, a unique token, or specific device claims) during the authentication flow. Entra ID Conditional Access can then be configured to check for these unique identifiers, ensuring that only requests originating from the trusted Prisma Browser application are allowed.

Why the other choices are incorrect:

• Option A (List of known egress IP addresses associated with Prisma Browser’s cloud proxy infrastructure) is incorrect: While Prisma Browser uses cloud proxies, relying solely on egress IP addresses is not a robust or granular method for confirming the specific client application. IP addresses can be shared, change, or be spoofed, and they do not uniquely identify the Prisma Browser application itself versus other applications that might use similar network paths.
• Option C (Certificate thumbprint of Prisma Browser’s secure workspace key used for session encryption) is incorrect: Certificate thumbprints are primarily used for verifying the authenticity of certificates and securing communication channels (e.g., SSL/TLS). While encryption is vital, a certificate thumbprint of a ‘secure workspace key’ is not the primary mechanism Entra ID Conditional Access uses to identify and enforce policies based on the specific client application. Conditional Access typically evaluates application IDs, device compliance, or specific application claims.
• Option D (GlobalProtect mobile application installed on the user’s endpoint) is incorrect: GlobalProtect is a VPN client for ZTNA and provides secure connectivity. However, the presence of GlobalProtect on an endpoint does not guarantee that the access request is originating from the Prisma Browser application. A user could still attempt to access the SaaS application using a standard web browser even if GlobalProtect is installed and running, thereby circumventing the Prisma Browser requirement. The question specifically asks to confirm the request originates from Prisma Browser.