The correct answer is B . In Zero Trust architecture, an initiator is not limited to a human user on a laptop. It can include many entity types that request access to a service, application, or data set. These can include managed devices, Internet of Things (IoT) systems, Operational Technology (OT) assets, and application workloads . This reflects the broader Zero Trust principle that trust decisions are applied to all requesting entities, not only to traditional employee endpoints.

This is important because modern enterprises no longer consist only of users on corporate desktops. They also include sensors, industrial systems, virtual machines, containers, and cloud-hosted workloads that generate access requests. Zero Trust must therefore evaluate the identity and context of these initiators using policy, posture, and risk rather than relying only on network location.
The other options are not correct because IP addresses, ports, and sockets are technical connection details, not the actual initiating entity in the Zero Trust model. A walled garden is also a network design concept, not a type of initiator. Therefore, the best answer is devices, IoT/OT, and workloads

The correct answer is C . In Zero Trust architecture, policy enforcement is not based on a single attribute such as identity, time, or location alone. Zscaler’s guidance states that policy decisions evaluate the entire user context , including the user, machine, location, group, and more . It also provides examples where the same user can be allowed or denied access depending on device posture , location, and other conditions.
The ZPA architecture similarly explains that access policy rules are built from application segments , SAML attributes , client types , and posture profiles , with additional context such as network location and device posture. That means effective policy enforcement depends on knowing the full access context : who the user is, what application is being requested, what device is being used, the posture of that device, and any other policy conditions tied to the request.
Options A, B, and D are each only partial inputs. Time of day, location, and verified identity can matter, but none of them alone is sufficient. The best and most complete answer is full context of the user, app, device posture, and related attributes.

The correct answer is D. Located anywhere and built on IPv4 or IPv6. In Zero Trust architecture, the network and application access model is not tied to a specific physical location, branch, or data center. Zscaler’s Zero Trust guidance emphasizes that users, devices, and applications can be securely connected in any location , which is a core shift away from legacy perimeter-based designs. The architecture is also described as IP independent , meaning policy and access decisions are not fundamentally anchored to traditional network constructs such as fixed addressing or trusted subnets. This is why Zero Trust can operate across modern environments regardless of where workloads reside.
The option about VPN concentrators is incorrect because VPN-based architecture is associated with legacy remote-access models that extend network trust and expose services differently from Zero Trust. In contrast, Zero Trust reduces implicit trust, avoids broad network-level access, and focuses on secure, application-aware connectivity. Therefore, the most complete and accurate answer is that a Zero Trust network can be located anywhere and built on IPv4 or IPv6 , rather than being limited to a legacy transport or perimeter model.

The correct answer is B . The core security risk of a split tunnel VPN is loss of visibility and consistent inspection for the traffic that bypasses the tunnel and goes directly to the internet. Zscaler’s Secure Mobile Access reference architecture explains that traditional VPNs backhaul traffic to a central data center for security through a legacy appliance stack, while modern remote work leads to a lack of visibility into what users are accessing and how the network is performing when the organization no longer controls the path.
ZIA guidance similarly states that user traffic must be forwarded to the nearest ZIA Service Edge so it can be inspected and either forwarded or blocked according to policy, and that the same authentication and policy should follow the user wherever they are. If some traffic exits directly to the internet outside that enforcement path, the organization loses the visibility and control needed to make reliable policy decisions on those flows. That is the real Zero Trust concern with split tunneling. It creates blind spots rather than a uniformly enforced security model. Therefore, the best answer is loss of visibility into traffic going directly to the internet

This statement is false. In Zscaler’s Zero Trust architecture, the recommended design objective is to

inspect as much encrypted traffic as possible because inspection enables security controls such as
malware protection, sandboxing, intrusion prevention system (IPS), browser isolation, Data Loss
Prevention (DLP), cloud application controls, tenancy restrictions, and file type controls. The
reference architecture states that inspecting all TLS/SSL traffic provides the fullest visibility and
strongest protection across the Zero Trust Exchange. However, the same document also clearly
confirms that inspection bypasses are supported in specific circumstances. These documented
exceptionsinclude banking and finance destinations, healthcare destinations, business functionsthat
require unencryptable traffic, certificate-pinned applications, and some Microsoft 365 application
flows that may not function properly under inspection. Zscaler strongly recommends using bypasses
only in extreme circumstances, but it does not say exceptions are architecturally impossible.
Therefore, from a verified Zero Trust design standpoint, full inspection isthe preferred security
posture, while selective exceptions are still an allowed and documented deployment option.

In Zero Trust architecture, policy enforcement is conditional and context-based, not limited to a

simple binary allow-or-block model. Zscaler’s reference architectures explain that policy is evaluated
using the full user context, including identity, device posture, location, group membership, and other
conditions. Access decisions are therefore based on whether specific policy conditions are true,
rather than only on static network attributes such as source IP address. For example, the same
authenticated user may be allowed access from a managed device at headquarters but denied from
an airport, even with the same credentials.
Zscaler documentation also shows that Zero Trust policy can go beyond simple pass or deny
outcomes by applying additional controls. In DNS Security and Control, requests can be allowed,
blocked, or modified. In ZIA policy development, Cloud App controls allow more granular outcomes
than standard allow/block,such as restricting specific actions, applying quotas, or controlling what a
user can do inside an application. This reflects the Zero Trust principle that enforcement is adaptive,
granular, and tied to business and security context rather than network location alone.

The correct answers are A, C, and D. In a legacy architecture, applications that are exposed and discoverable on the public Internet are usually protected by building a DMZ (demilitarized zone) and placing multiple security technologies in front of the service. This commonly includes a large security stack made up of separate appliances or services for functions such as load balancing, firewalling, distributed denial-of-service (DDoS) protection, and related edge security controls. A web application firewall (WAF) is also a standard protective element in these public-facing designs because it adds inspection and protection for web-based attack patterns and internet-originated abuse. Option B, DAST, is not a correct answer because Dynamic Application Security Testing is a testing and assessment method, not a live architectural protection control that sits inline to defend exposed services in production. Zero Trust architecture contrasts with this legacy model by removing direct public discoverability and reducing dependence on a complex exposed edge stack. Instead of defending openly exposed applications with layered perimeter tools, Zero Trust aims to make applications less discoverable and access more identity- and policy-driven.