Few security terms have been stretched as far as “zero trust”. Vendors attach it to firewalls, VPN replacements, identity platforms and endpoint agents. Underneath the marketing is a fairly precise idea, and the best plain statement of it comes from the US National Institute of Standards and Technology in Special Publication 800-207, published in 2020. The term itself was popularised earlier by John Kindervag at Forrester, and Google’s BeyondCorp programme was an influential early implementation.

The problem it responds to

Traditional network security drew a perimeter. Inside the office network was trusted; outside was not. Once you were inside, through the office Ethernet or the VPN, you could reach a lot.

That model weakened for several reasons. Applications moved to SaaS and public cloud, outside the perimeter. Employees work from homes, client sites and cafés. Attackers learned that phishing one employee puts them inside the trusted zone, after which internal systems often trust them by default. The perimeter became both porous and less relevant.

Zero trust drops the assumption that network location implies trust. A request from a laptop on the office LAN is evaluated the same way as one from a home broadband connection in Nagpur.

The core tenets, paraphrased

SP 800-207 lists seven tenets. In plain terms:

  1. All data sources and computing services are treated as resources to be protected, including SaaS and personal devices that access company data.
  2. All communication is secured regardless of network location. Internal traffic gets the same protection as external.
  3. Access is granted per session, to an individual resource. Access to one application does not imply access to another.
  4. Access is decided by dynamic policy using identity, device state, behaviour and environmental attributes.
  5. The organisation monitors and measures the integrity and security posture of all owned and associated assets.
  6. Authentication and authorisation are dynamic and strictly enforced before access is allowed, and re-evaluated during use.
  7. The organisation collects as much information as possible about assets, network traffic and access requests, and uses it to improve its security posture.

The moving parts

The document describes a logical architecture with three main components. A Policy Engine makes the decision to grant, deny or revoke access. A Policy Administrator acts on that decision, for example by issuing a session token or telling the gateway to open a path. Together these form the policy decision point. A Policy Enforcement Point sits in front of the resource and allows or blocks the actual connection.

The policy engine draws on inputs such as the identity provider, device management and EDR status, threat intelligence, activity logs and data classification. A concrete request might go like this. Priya, a finance analyst, opens the expense approval application from her laptop. The enforcement point, perhaps an identity-aware proxy, holds the request. The policy engine checks that she authenticated recently with phishing-resistant MFA, that her laptop is company-managed with disk encryption on and the EDR agent healthy, and that she is in the finance group. All true: access is granted to that one application. If her EDR agent later reports malware, the session can be revoked mid-use.

Myths worth retiring

“Zero trust means trusting nothing.” It means trusting nothing implicitly. Trust is still granted, but explicitly, per request, based on evidence, and for limited scope and time.

“We bought a ZTNA product, so we are zero trust.” Zero trust network access tools are a useful enforcement point, often replacing broad VPN access with per-application access. But if identities are weak, devices unmanaged and policies permissive, the product enforces a poor policy very efficiently. NIST itself frames zero trust as a journey that most organisations will reach incrementally, running hybrid environments for a long time.

“The network no longer matters.” Network controls still matter. Segmentation limits damage, and encryption protects data in transit. What changes is that network location stops being the main reason to grant access.

“Zero trust is only for large enterprises.” Many smaller companies are already partly there without calling it that: their applications are SaaS, their staff log in through a single identity provider with MFA, and devices are managed. Adding device compliance checks to conditional access policies is a meaningful zero trust step that a 50-person company can take.

A practical order of work

Because zero trust decisions depend on identity and device signals, those foundations come first. A realistic sequence for most organisations:

  • Consolidate identities into one provider with strong MFA, and connect applications to it through single sign-on.
  • Build an inventory of devices and manage them, so device health can be checked.
  • Define policies for the most sensitive applications first: who, from what kind of device, under what conditions.
  • Replace broad network access, such as full VPN access to the data centre, with per-application access for those systems.
  • Feed logs from every decision into monitoring, and tighten policies based on what you learn.

Where it gets hard

Legacy applications that cannot use modern authentication are the most common obstacle. An old thick-client ERP talking directly to a database over a proprietary protocol cannot easily sit behind an identity-aware proxy. These often end up in tightly segmented enclaves reached through hardened jump hosts, a pragmatic compromise.

Machine-to-machine traffic is another challenge. Services calling services need identities too, through workload identity, mutual TLS or short-lived credentials, and that is engineering work rather than a configuration change. Finally, the policy engine becomes critical infrastructure. If the identity provider is unavailable, nobody can access anything, so its resilience and its own security deserve as much attention as any application it protects.

Written by the RP Strativa Team. The employee and scenario in the example are illustrative.

Leave a Reply

Trending