Once an administrator has moved away from Prisma Access ' s default, automatic bandwidth-sharing behavior and explicitly customized bandwidth allocation per site on a shared IPSec termination node, the platform respects that deliberate, manual configuration rather than silently recalculating or redistributing percentages whenever a new site is added to the same node. Onboarding a fifth branch office onto a termination node where the existing four sites already have customized, fixed bandwidth values does not trigger an automatic rebalancing to a new even split; instead, the new site simply has no bandwidth allocation defined for it and will remain unallocated, effectively receiving no guaranteed or prioritized QoS treatment, until the engineer explicitly assigns it a bandwidth value as part of onboarding. This makes option B the accurate description of default platform behavior. Options A and C both describe an automatic, evenly-redistributed percentage outcome (25% and 20% respectively, which would correspond to five equal shares or four equal shares) that does not reflect how customized QoS interacts with new site onboarding - automatic even redistribution is the behavior only when no manual customization has been introduced in the first place, and once customization exists, the platform does not silently override or reflow it. Option D is incorrect because new branch offices absolutely can be added to an IPSec termination node with existing customized QoS; the addition itself is fully supported, it simply requires the administrator to manually define that site ' s bandwidth.
Reference:Prisma Access Remote Networks - QoS Bandwidth Allocation per IPSec Termination Node.

The question specifically asks for components that provide data center connectivity over the internet, which is the distinguishing factor between the four options. Service connections are the classic IPSec-based method:
they build an encrypted tunnel from the customer ' s data center edge device to Prisma Access across the public internet, requiring no private circuit. ZTNA Connector achieves the same outcome through a different architecture - a lightweight, outbound-only connector deployed in the data center that establishes a secure, brokered tunnel to the nearest Prisma Access cloud gateway, again entirely over the internet, without requiring inbound firewall rules or a traditional IPSec peer. Both are therefore valid answers. Colo-Connect is explicitly excluded because it is built for the opposite use case: it delivers private, high-bandwidth connectivity to data centers using GCP Dedicated or Partner Interconnects, bypassing the public internet entirely to achieve up to 100 Gbps with lower latency and jitter - the architecture exists specifically because customers want to avoid internet transport for their highest-throughput sites. SD-WAN Connector is not a genuine Prisma Access private-application access method; Prisma SD-WAN integrates with Prisma Access through ION devices acting as CPE for remote networks or service connections, not as a distinct " connector " component in this context, so it does not belong in this answer set.
Reference:Prisma Access Service Connections, ZTNA Connector, and Colo-Connect - Private Application Access Methods.

IKE and IPSec negotiation events - including successful and failed Phase 1 (IKE SA) and Phase 2 (IPSec SA) exchanges, proposal mismatches, and negotiation timeouts - are recorded by PAN-OS as System log entries, not as part of the Traffic, Tunnel, or Decrypt log facilities, which each capture a different category of information. Because Phase 1 has already completed successfully in this scenario but Phase 2 negotiation appears to be failing or stalling, the actual diagnostic detail explaining why - such as a proxy-ID/traffic- selector mismatch, an unsupported Phase 2 encryption or authentication algorithm, or a PFS group mismatch between the CPE and Prisma Access - will be recorded as a specific IKE/IPSec negotiation message in System logs, making option B the correct log facility to review. Decrypt logs (option A) capture SSL/TLS decryption events for inspected web traffic and have no relevance to IPSec tunnel negotiation, which is a separate control-plane process entirely. Traffic logs (option C) record session-level information for traffic that has already been successfully permitted through a completed policy match; since the tunnel ' s data plane is not yet fully established, there is no session traffic to log in the first place. Tunnel logs (option D) generally reflect the operational status and utilization of an already-established tunnel, not the underlying IKE/IPSec negotiation failure detail needed to diagnose why Phase 2 never completed.
Reference:PAN-OS/Strata Logging Service - System Logs for IKE Phase 1/Phase 2 Negotiation Troubleshooting.

Panorama ' s multitenancy implementation for Prisma Access relies on Access Domains as the primary boundary mechanism: when a tenant is created, Panorama automatically generates the device groups, templates, and template stack associated with that tenant and binds them to a dedicated access domain.
Restricting an administrator to that access domain confines their visibility and configuration rights strictly to the objects belonging to that tenant, which is exactly the outcome the question requires - full access within the tenant, no visibility into any other tenant ' s device groups or templates. This makes option A the structurally correct answer, because the access domain is the object that actually enforces the tenant boundary; a custom role alone, without an access domain restriction, defines what privileges an administrator has but not which tenant ' s objects those privileges apply to. Options B and C describe custom administrative roles, which are a necessary complement to access domains for fine-tuning specific privilege sets, but neither role definition by itself creates the tenant isolation the scenario demands - a role with " all privileges " or with device-group/template privileges could still be applied across every tenant ' s device groups unless paired with an access domain restriction. Assigning the Superuser role (option D) is explicitly the wrong direction:
Superuser grants unrestricted access across the entire Panorama instance and all tenants, which directly violates the requirement to restrict access to other tenants.
Reference:Prisma Access Multi-Tenancy (Panorama) - Access Domains and Tenant-Level Administrative Roles.

Prisma Access ' s automatic gateway selection logic follows a defined fallback hierarchy specifically designed to keep mobile users connected even when they travel to a country without an onboarded, in-country Prisma Access location. If a user cannot connect to an in-country location, the GlobalProtect app first attempts a regional fallback location - a nearby, same-theater location the organization has onboarded (for example, users elsewhere in Asia, Australia, and Japan falling back to a regional hub such as Hong Kong, Singapore, or Japan Central) - which keeps latency reasonable by staying within the same broad geography. If no suitable regional location is available or reachable, the client falls further back to one of a small, fixed set of global fallback locations (including Hong Kong, Netherlands Central, and US Northwest) that are specifically designated to accept client connections from anywhere in the world, guaranteeing a connection path of last resort regardless of where the traveling user is located. This two-tiered regional-then-global fallback behavior is exactly what makes options B and C the correct pair. " Backup " (option A) is not the term used for this automatic gateway-selection fallback behavior in GlobalProtect ' s Prisma Access location logic. " Local zone
" (option D) does not describe a fallback gateway category at all - it is not part of the documented regional
/global fallback location terminology and does not apply to a traveling user with no in-country location available.
Reference:GlobalProtect - How the App Selects Prisma Access Locations for Mobile Users (Regional and Global Fallback).