SD-WAN-Engineer Palo Alto Networks SD-WAN Engineer Question and Answers

Comprehensive and Detailed Explanation

The Prisma SD-WAN (ION) QoS engine utilizes a hierarchical queuing structure designed to provide granular control over application performance. Each WAN interface on an ION device supports a total of 16 QoS queues.

This 16-queue structure is derived from a matrix of 4 Classes (often referred to as Priority Classes) multiplied by 4 Application Criteria (Traffic Types).2

4 Priority Classes: The system defines four high-level business priority categories:3

Platinum (Highest priority)4

Gold

Silver

Bronze (Lowest priority/Best Effort)5

4 Application Criteria (Sub-queues): Within each of the four priority classes, the system further categorizes traffic into four specific application types to ensure proper handling (e.g., ensuring voice doesn't get stuck behind bulk data even within the same priority level):6

Real-Time Video

Real-Time Audio

Transactional

Bulk7

Calculation: 4 Priority Classes × 4 Application Types = 16 Total Queues per interface. This structure allows the scheduler to ensure that a "Platinum" voice call is prioritized over "Platinum" bulk data, and both are prioritized over "Gold" traffic.

Comprehensive and Detailed Explanation

In Prisma SD-WAN Path Policies, the Service & DC Group (Destination) field determines where the traffic is sent.

Direct: This is the specific keyword/object used to instruct the ION to route traffic directly out to the local WAN interface (Local Breakout) towards the Internet, without encapsulation in a VPN tunnel. This is the correct setting for Guest Wi-Fi, SaaS applications (like Office 365), or any public web browsing that does not need to be backhauled.

Standard VPN / Default-Cluster: These options direct traffic into an IPSec overlay tunnel destined for a Data Center or another ION. Selecting these would "backhaul" the guest traffic, which contradicts the requirement for DIA.

When "Direct" is selected, the ION uses its available "Internet" category links. The policy can further specify which internet link to use (e.g., "Use Broadband, avoid LTE") via the path preference list, but the Destination type must be "Direct".

Prisma SD-WAN High Availability (HA) for branch ION devices, particularly the Gen-2 ION 1200-S, is designed to provide "100% WAN Capacity" preservation during a hardware or power failure. This is achieved through the use of Bypass Pairs (Fail-to-Wire). In the provided topology, the ISP A and LTE/5G circuits are cross-connected using the bypass ports (typically ports 3 and 4 on the ION 1200-S).

When the "Active" ION device loses power, the internal physical relays in its bypass ports transition to a closed state, effectively creating a physical bridge between the ports. In this scenario, the LTE/5G signal—which enters the Active ION's port 4—is mechanically bridged to port 3, allowing it to pass through to port 4 of the Standby ION. Simultaneously, ISP A is already connected to the Standby ION. Consequently, once the Standby device completes its transition to the "Active" state, it has physical access to both WAN circuits, validating Statement A.

Regarding the LAN transition, Prisma SD-WAN does not use standard VRRP for ION-to-ION HA; instead, it uses a proprietary Control Plane HA mechanism. When the failover occurs, the newly active ION takes over the IP addresses of all configured Switch Virtual Interfaces (SVIs) and LAN interfaces. To ensure the downstream Layer 2 infrastructure (like the LAN switches shown in the diagram) updates its MAC address tables to point to the new physical hardware for those IPs, the newly active ION immediately broadcasts a Gratuitous ARP (GARP). This ensures that LAN traffic is correctly steered to the new device without a significant timeout, validating Statement C.

In Palo Alto Networks PAN-OS SD-WAN environments, automation and orchestration are key components for service providers managing large-scale deployments. The PAN-OS REST API provides a modern, structured way to programmatically manage configuration objects, including those required for SD-WAN functionality.

When an application is designed to push changes directly to devices (individual firewalls) rather than through a centralized template in Panorama, it must interact with the firewall's local REST API. To successfully create a virtual SD-WAN interface, the application must target the correct resource URI. In the PAN-OS API schema, the logical SD-WAN interface—which groups physical links to enable application-based path selection—is managed via the sdwanInterfaces parameter within the REST API.

It is important to distinguish between the interface itself and the profiles that support it. Option A refers to sdwanInterfaceprofiles, which are the objects used to define the characteristics of a link (such as bandwidth, link type, and monitoring frequency), but not the interface itself. Furthermore, since the scenario specifies making changes "directly to devices," the target must be the firewall rather than Panorama. While Panorama can manage these objects via templates, a direct-to-device automation workflow necessitates using the firewall’s REST API endpoint. Utilizing the REST API over the legacy XML API is the recommended standard for modern integrations due to its ease of use with JSON payloads and alignment with contemporary DevSecOps practices. By using the sdwanInterfaces parameter on the firewall, the MSP application can programmatically bind physical Layer 3 interfaces to the SD-WAN fabric.

Comprehensive and Detailed Explanation

In a Prisma SD-WAN Data Center deployment, the operational state of the Secure Fabric VPNs (overlay tunnels) is directly tied to the health of the BGP Core Peer configuration.4

Core Peer Dependency: DC ION devices typically peer with the data center core switch (Core Router) via BGP to learn the subnets (prefixes) for the applications hosted in the DC. The Prisma SD-WAN controller monitors this BGP peering status.5

Controller Logic: If the BGP Core Peer on a DC ION goes down (or is not established), the controller automatically marks the VPN tunnels terminating at that specific ION as "Inactive".6 This is a fail-safe mechanism designed to prevent remote branches from sending traffic to a DC ION that has lost conne7ctivity to the internal data center network (and thus the applications).

Scenario Analysis: In this scenario, DC ION-1 has active VPNs, meaning its BGP Core Peer is UP and it is successfully advertising reachability. DC ION-2 has no active VPNs, which strongly indicates that its BGP Core Peer is down.8 Because the controller sees the peer is down, it suppresses the tunnel establishment or marks existing tunnels as inactive to ensure traffic is only directed to the healthy node (ION-1).

Comprehensive and Detailed Explanation

This behavior describes the "Best Available Path" logic inherent in Prisma SD-WAN's availability design.

SLA Thresholds: Path Quality Profiles act as filters to identify compliant paths.

Total Violation: If all configured "Active" paths violate the SLA (e.g., Path A has 2% loss, Path B has 5% loss, and the threshold is 1%), the system does not drop the traffic (Option A) because maintaining connectivity is prioritized over perfect quality.

Selection Logic: The system enters a fallback state where it compares the available active paths and selects the "Least Bad" one—the path that is closest to meeting the SLA (in this case, Path A with 2% loss).

Backup Paths: Traffic would only move to a Backup path (Option D) if the policy explicitly configures the backup path to engage upon SLA violation of the active set. However, strictly speaking, if only active paths are considered and all fail, it picks the best of the active group rather than blackholing the traffic.

Comprehensive and Detailed Explanation

In the Prisma SD-WAN Flow Browser, the Flow State provides a real-time snapshot of the TCP/UDP session lifecycle.

INIT (Initialization): This state indicates that the ION device has seen the initial packet of a new session (typically a TCP SYN) originating from the client (Source), but it has not yet seen a return packet (such as a TCP SYN-ACK) from the destination server.

Diagnosis: A flow stuck in INIT is a classic indicator of a "Blackhole" or reachability issue downstream. It implies that the ION successfully routed the packet out toward the destination, but the destination did not reply. Common causes include:

The server is offline.

A firewall in the path (or on the server itself) is dropping the traffic.

Routing is broken on the return path (asymmetric routing where the return traffic bypasses the ION).

If the flow had been denied by the ION's own firewall (Option C), the state would typically show as DENY or REJECT. If the handshake completed (Option A), the state would be ESTABLISHED. Therefore, INIT points to a lack of response from the remote end.

In a Prisma SD-WAN deployment, the routing of traffic between branches and Data Centers (DCs) relies on the proper synchronization between the AppFabric (the overlay) and the local routing protocols (the underlay/LAN side). In this scenario, the branch can successfully reach DC1, indicating the branch ION is correctly participating in the fabric. However, traffic to DC2 (10.2.2.22) is failing.

The DC2 site has the site prefix 10.2.2.0/23 configured. In Prisma SD-WAN, defining a site prefix informs the Controller that this specific subnet "belongs" to that site, causing the Controller to advertise reachability for this prefix to all other ION devices in the fabric. Consequently, when the branch ION (192.168.1.123) attempts to reach 10.2.2.22, it correctly identifies DC2 as the destination and encapsulates the traffic toward the DC2 ION.

The bottleneck occurs once the packet arrives at the DC2 ION. While the ION is advertising the branch subnet (192.168.1.0/24) to the DC Core (ensuring the return path), the ION itself must know how to forward the incoming traffic from the branch to the internal DC network. If the DC2 ION does not have a specific route in its local routing table for the 10.2.2.0/23 subnet pointing to the DC Core's internal interface, the packet will be dropped.

According to Palo Alto Networks best practices for Data Center ION deployment, a static default route (0.0.0.0/0) should be configured on the ION device pointing toward the DC Core’s next-hop IP address. This ensures that any traffic received from the AppFabric destined for internal DC resources—which are not directly connected to the ION—is successfully handed off to the core switching fabric for final delivery. Adding this default route (Option A) resolves the reachability issue by providing the "last-hop" routing instruction within the DC.

Comprehensive and Detailed Explanation

In a Prisma SD-WAN High Availability (HA) deployment, the HA Control Interface is the critical lifeline used to synchronize state, heartbeats, and flow information between the Active and Standby ION devices.

The strict requirement for this connection is that it must be Layer 2 adjacent.

Best Practice: A direct physical cable connection between the designated HA ports of the two devices (e.g., Port 2 on Device A to Port 2 on Device B).

Alternative: Connectivity through a switch on a dedicated, isolated VLAN is supported, provided the devices are in the same broadcast domain and subnet.

Routing (Layer 3) is not supported for the HA Control link because the keepalive mechanism relies on low-latency, multicast/broadcast-level adjacency to detect failures instantly (sub-second failover). If the HA link were routed (Option A), network latency or router convergence issues could cause "Split-Brain" scenarios where both devices assume the Active role, leading to IP conflicts and traffic loops. Option C is incorrect because the Controller is too slow to manage real-time failover; the decision must be local.

Comprehensive and Detailed Explanation

In the Prisma SD-WAN (CloudGenix) Zero Touch Provisioning (ZTP) lifecycle, the device status transitions through specific stages that indicate its readiness and connectivity.

When an administrator enters the Claim Code (or Serial Number/Claim Code pair) into the portal, the device status immediately updates to "Claimed".

This status confirms that the portal has registered the device's unique identity and associated it with the customer's tenant. However, "Claimed" does not necessarily mean the device is fully operational or passing traffic yet. It simply signifies that the ownership is verified.

Once the physical device at the site successfully connects to the internet and reaches the Prisma SD-WAN Controller (using the call-home function), it will authenticate using its installed certificate. Upon successful authentication and the establishment of the secure control channel, the status will transition from "Claimed" to "Online".

Only after the device is "Online" can the controller push the specific site configuration (Device Shell), policies, and IP addressing required for the device to become "Provisioned" and eventually "Active" in the data path. If the device remains in the "Claimed" state for an extended period, it indicates that the hardware has not yet successfully contacted the controller, which prompts troubleshooting of the physical internet circuit or firewall rules upstream.

Comprehensive and Detailed Explanation

Prisma Access manages Remote Network bandwidth using an Aggregate Bandwidth licensing model.

Compute Locations: When you purchase bandwidth (e.g., 1 Gbps), you allocate it to specific Prisma Access Compute Locations (e.g., US West, Europe Central).

Shared Pool: All branch sites (Remote Networks) that connect to that specific Compute Location share the allocated bandwidth pool. For example, if you allocate 500 Mbps to "US West" and connect 10 branches to it, they compete for that 500 Mbps aggregate.

Bursting: An individual branch is not strictly rate-limited to a "slice" (e.g., 50 Mbps) unless you explicitly configure QoS guarantees. By default, a single branch can burst and consume a large portion of the aggregate pool if other branches are idle. The enforcement happens at the Region/Compute Node level, ensuring the total throughput does not exceed the licensed capacity for that region.

Comprehensive and Detailed Explanation

The RMA replacement process in Prisma SD-WAN is designed to be seamless, leveraging the decoupling of logical configuration from physical hardware.

Replace Device Workflow: The administrator should use the "Replace Device" (or RMA) function within the portal. This workflow allows you to select the "Defective" device (old serial) and the "Replacement" device (new serial).

Configuration Transfer: Once executed, the system automatically binds the existing Device Shell (which contains all interface configs, routing policies, and site associations) to the new hardware's serial number. The new device, once connected to the internet, will "call home," identify itself, and download the exact configuration of the previous unit.

License Transfer: While the configuration moves automatically, the Support License transfer typically requires a specific step in the Customer Support Portal (CSP) or happens automatically if processed as a formal RMA order. Options A and D are incorrect because they involve manual reconfiguration, which is unnecessary and error-prone. Option C is incorrect as the ION platform relies on cloud-based config management, not local USB backups for hardware swaps.

In Prisma SD-WAN, the effective throughput for any given circuit is fundamentally dictated by the Circuit Configuration defined at the site level. When a branch experiences a "throughput ceiling" (e.g., traffic capped at 200 Mbps on a 1 Gbps physical link), the most likely cause is that the software-defined bandwidth limit for that circuit has been set incorrectly in the Prisma SD-WAN Controller.

Prisma SD-WAN ION devices do not simply forward traffic at the maximum physical line rate by default; they rely on the administrator-defined Upstream and Downstream bandwidth values to perform traffic shaping, policing, and path selection. If a circuit is physically capable of 1 Gbps but is configured in the portal as having only 200 Mbps, the ION device will enforce this 200 Mbps limit to prevent oversubscribing the link and to ensure that Quality of Service (QoS) and path selection calculations remain accurate based on the assumed capacity.

To resolve this, an engineer must navigate to the Site Configuration, locate the specific WAN circuit, and verify that the bandwidth settings match the actual service provider's handoff. If these values are set lower than the actual link speed, the device will artificially throttle the traffic. While ensuring the WAN interface is set to the correct speed/duplex (Option B) is a valid physical layer check, and QoS/Performance policies (Options A and C) manage how that bandwidth is used, it is the Circuit Configuration that defines the total available bandwidth for the SD-WAN fabric to utilize. Correcting this configuration allows the ION device to scale its throughput to match the full 1 Gbps capability of the broadband connection.

For a massive rollout involving 1,000 branch sites, Prisma SD-WAN (formerly CloudGenix) provides a specialized workflow known as Bulk Site Configuration. This method is designed to minimize manual intervention and maximize deployment velocity by leveraging Site Templates and Device Shells.

In this scenario, the primary architectural advantage of Option C is the use of Pre-Staging. By exporting an empty SD-WAN device CSV from the Prisma SD-WAN Controller and populating it with data from the corporate CMDB, administrators can perform a bulk upload to create hundreds or thousands of sites and device shells simultaneously in the management portal. A "Device Shell" acts as a placeholder for a physical ION device that has not yet connected to the cloud. It contains all the site-specific configuration—such as interface roles, circuit labels, and IP addressing—waiting for a serial number to be associated with it.

When the field technician performs the physical "rack and stack," they simply connect the ION device to the internet (via ISP modem or cellular). Through Zero Touch Provisioning (ZTP), the device automatically "phones home" to the Prisma SD-WAN Cloud Controller using its Manufacturer Installed Certificate (MIC). Because the configuration was pre-created via the CSV bulk upload, the controller recognizes the device (once assigned to its shell) and immediately pushes the complete configuration. This eliminates the need for the field tech to access a console port or perform local configuration, reducing their on-site time to the bare minimum. While APIs (Option D) can be used for automation, the built-in CSV template workflow is the standard, documented "best practice" for rapidly translating CMDB data into a functioning SD-WAN fabric at this scale.

The provided graph displays the Signal-to-Noise Ratio (SNR) for two cellular carriers, Carrier-1 (blue line) and Carrier-2 (green line), over a specific period. In cellular communications, SNR is a critical metric used to determine the quality of a wireless signal. A higher SNR indicates a cleaner, stronger signal, while a lower SNR indicates that the signal is being "drowned out" by background noise or interference, which directly correlates to performance degradation, packet loss, and lower throughput.

Looking at the graph, Carrier-1 experiences a significant and sustained drop in SNR, falling from roughly 4.5 dB to nearly 0.5 dB for the majority of the time duration. This drastic reduction in signal quality strongly suggests that Carrier-1 may have experienced performance degradation (Option A). During this dip, the link quality would likely fall below the configured thresholds for business-critical application traffic.

Because Prisma SD-WAN is an application-defined fabric that continuously monitors path health, the ION device would detect this degradation on Carrier-1. If Carrier-2 maintains a significantly higher and more stable SNR (as shown by the green line remaining between 4.5 dB and 6.5 dB), the ION device's Path Selection engine would automatically steer traffic away from the degraded link. Consequently, it is highly probable that Carrier-1 traffic switched over to Carrier-2 (Option D) to maintain the application SLA. This automated failover is a core strength of the Prisma SD-WAN architecture, ensuring that the best available path is utilized based on real-time link statistics rather than simple "up/down" states.

Comprehensive and Detailed Explanation

To defend against volumetric attacks such as TCP SYN Floods, UDP Floods, or ICMP Floods, Prisma SD-WAN (like PAN-OS) utilizes Zone Protection Profiles.

Function: A Zone Protection Profile is a specific security object designed to screen traffic for protocol anomalies and flood behaviors before it is processed by the complex firewall policy engine. It sets thresholds (e.g., "Max 1000 SYNs/sec"). If the traffic rate exceeds this threshold, the system triggers an action (Alarm, Drop, or SYN Cookies) to protect the device's resources.

Application: Unlike a standard ZBFW Rule (A) which filters based on Source/Destination/App-ID (which might still allow the initial handshake packets that cause the flood), a Zone Protection Profile is applied to the Zone object itself (in this case, the LAN Zone). This ensures that the flood is mitigated at the ingress stage, preventing the ION's session table and CPU from being exhausted by the attack.

Comprehensive and Detailed Explanation

The transition from "Claimed" to "Online" depends entirely on the ION device's ability to establish a secure, persistent management tunnel to the Prisma SD-WAN Controller.

Connectivity Requirements: The ION device initiates an outbound connection to the controller on TCP Port 443 (HTTPS). It also requires accurate time synchronization to validate SSL certificates, necessitating access to NTP (UDP Port 123).

Scenario Analysis: Since the installer can browse the internet from the LAN, we know the physical link and basic routing/NAT are functional. The issue is specific to the management plane traffic.

Root Cause: If an upstream firewall (e.g., a corporate edge firewall or ISP filter) is inspecting SSL traffic or blocking specific FQDNs/Ports required by the ION, the device cannot complete the handshake. Consequently, it remains "Claimed" (registered in the database) but cannot go "Online" (active management session). Options A, C, and D prevent provisioning (configuration push) but generally do not prevent the device from initially checking in and going "Online" if the pipe is open.

Prisma SD-WAN is built on an application-defined fabric that prioritizes deep visibility into network traffic and application performance.1 To deliver the high-fidelity flow data and application visibility required for modern operations, Prisma SD-WAN utilizes IPFIX (Internet Protocol Flow Information Export).2 IPFIX is a standardized protocol based on NetFlow v9 that allows for the export of IP flow information from network devices to a collector or management system.3

In the Prisma SD-WAN architecture, ION devices act as the exporters.4 Because the system is application-aware, it doesn't just export basic 5-tuple information (source/destination IP, ports, and protocol); it exports rich metadata including application IDs, performance metrics (latency, jitter, packet loss), and path information. This allows the Prisma SD-WAN Controller and the associated Analytics engine to reconstruct a complete picture of every flow in the network.

While other protocols like SNMPv3 are supported for basic device health monitoring (such as CPU or interface status) and ADEM (Autonomous Digital Experience Management) provides end-to-end visibility for mobile users or SASE-connected branches, IPFIX is the primary "engine" for flow-level data across the SD-WAN fabric. Unlike traditional IP SLA, which relies on synthetic probes, the IPFIX-based monitoring in Prisma SD-WAN uses real-time application traffic to assess performance. This ensures that the visibility provided in the Flow Browser and Analytics dashboards accurately reflects the actual user experience, enabling granular troubleshooting and proactive capacity planning.

Comprehensive and Detailed Explanation

Prisma SD-WAN (formerly CloudGenix) defines three distinct Operational Modes for a branch site, which determine how the ION device processes traffic and interacts with the network.

Analytics Mode (Monitor): In this mode, the ION device is typically deployed inline or in a "promiscuous" monitor state to gain visibility into network traffic without actively enforcing path selection policies.1 It "learns" applications, bandwidth usage, and network characteristics (auditing) but does not steer traffic or block flows.2 This is often used during Proof of Concepts (POVs) or the initial "burn-in" phase of a deployment to generate reports without risking network disruption.

Control Mode: This is the full production state. In Control Mode, the ION device actively enforces Path Policies, QoS Policies, and Security Policies. It builds Secure Fabric VPN tunnels, steers traffic based on application SLAs (e.g., sending voice over MPLS and bulk data over Broadband), and handles failover events.3 This is the required mode for a fully functional SD-WAN site.

Disabled Mode: This mode effectively shuts down the site's SD-WAN functionality from the controller's perspective. It is an administrative state used when a site is being decommissioned, provisioned but not yet live, or isolated for troubleshooting. In this state, the device does not participate in the fabric.

In complex retail or banking environments, maintaining strict network segmentation is a regulatory and security requirement. Prisma SD-WAN utilizes Virtual Routing and Forwarding (VRF) to provide this isolation, ensuring that high-security traffic, such as ATM transactions or teller cash registers, remains logically separated from Guest Wi-Fi or general corporate applications. While isolation is the default state, route leaking is used to allow specific communication between these VRFs—for instance, allowing multiple isolated segments to reach a common shared service like a DNS server or a centralized security gateway.

To verify that these configurations have been correctly pushed from the Controller to the local ION device, administrators utilize the ION CLI (Command Line Interface) for deep-dive diagnostics. The command inspect vrf route_leak_rule all is the definitive tool for this purpose. Unlike "show" commands which typically provide interface status, "inspect" commands in the Prisma SD-WAN ecosystem are designed to pull real-time operational state data from the control plane's internal databases.

When executed, this command displays the specific prefix-level rules that allow routes to "leak" from one VRF table into another. It provides visibility into the source VRF, the destination VRF, and the exact network prefixes or default routes being shared. This is critical for troubleshooting "Day 2" operations; if a teller register cannot reach a shared database, the administrator can use this command to confirm if the necessary route leak rule is active and accurately reflecting the intent of the VRF Profile configured in the portal. Without this command, verifying inter-VRF reachability would be limited to trial-and-error connectivity tests, making it an essential part of the Prisma SD-WAN engineer's toolkit.

Prisma SD-WAN licensing is structured to provide flexibility while ensuring that all components of the secure fabric are correctly accounted for. To meet the requirements of this organization, we must calculate the necessary subscriptions for both the data center hubs and the distributed branch network.

First, we address the Data Center Subscriptions. The organization has two main geographically separate data centers and one small data center, all of which require High Availability (HA). In a Prisma SD-WAN deployment, HA at a site is achieved by deploying two ION devices in a cluster. Palo Alto Networks licensing requires a separate Data Center subscription for each ION device acting as a hub. Therefore, with three data center locations (2 main + 1 small) each requiring an HA pair (2 devices per site), a total of six data center subscriptions (Option A) are required to license all six hub appliances.

Second, we address the Branch Subscriptions. The organization has 50 branches but lacks accurate measurements of actual bandwidth consumption. Palo Alto Networks' best practice for such scenarios is the Aggregate Bandwidth Subscription model (Option B). Instead of purchasing a fixed "Branch subscription per site" (Option D)—which requires knowing the exact throughput needs for every individual location—the aggregate model allows the customer to purchase a total pool of bandwidth (e.g., 5 Gbps) that is shared across all 50 branch sites.

This "pay-as-you-grow" approach is ideal when consumption patterns are unknown or inconsistent. As branches utilize the bandwidth, it is deducted from the central pool. This avoids the risk of over-provisioning licenses at low-usage sites or under-provisioning at high-usage sites. Together, the six DC subscriptions and the aggregate bandwidth pool provide a fully licensed, HA-capable SD-WAN environment that aligns with Palo Alto Networks' scaling recommendations.

In Prisma SD-WAN, the Signal-to-Noise Ratio (SNR) is a critical metric used to monitor the health and performance of cellular WAN interfaces. SNR measures the strength of the desired signal relative to the background noise level; higher values indicate a cleaner signal, while lower values suggest that noise is overwhelming the signal, typically leading to increased packet loss, high latency, and reduced throughput.

Analyzing the provided graph, Carrier-1 (blue line) shows a severe drop in SNR, plummeting from approximately 4.5 dB to nearly 0.3 dB between 15:00 and 23:00. An SNR value this low is indicative of a failing or highly unstable link that cannot reliably sustain data traffic, directly supporting Inference A—that Carrier-1 experienced significant performance degradation. In contrast, Carrier-2 (green line) maintains a much higher and more consistent SNR throughout the same period.

Prisma SD-WAN’s AppFabric uses application-based path selection and SLA monitoring to ensure the best possible user experience. When the system detects that a primary path (like Carrier-1) has degraded below acceptable thresholds—often triggered by high loss or latency resulting from poor signal quality—it will dynamically steer application flows to an alternative healthy path. Therefore, Inference D is correct: because Carrier-1’s quality became untenable while Carrier-2 remained stable, the ION device would have likely initiated a path switchover to move traffic from the degraded Carrier-1 to the healthier Carrier-2.