Zero Trust is the principle. ZTNA is the implementation. SASE is the architecture that operationalises it at scale.

SASE, VPN and Zero Trust are often discussed as if they are interchangeable. They are not. For most Australian organisations, the right question is not “Should we buy SASE?” It is whether the environment is ready to move from network-level remote access to identity-led, application-level access — and if so, whether that starts with hardened VPN, ZTNA, SSE or full SASE.

See what decision you are actually making
This comparison covers

VPN, ZTNA, Zero Trust and SASE definitions; where VPN still fits; when ZTNA should replace user-facing VPN; when full SASE is justified; identity and application prerequisites; Essential Eight and cyber insurance implications; migration sequence and parallel operation; and Microsoft Entra SSE, Fortinet FortiSASE and combined architecture fit. Written for Australian organisations weighing remote access and secure network architecture decisions.

Decision framework

The right architecture depends on what the environment can support now.

The common mistake is selecting an architecture based on what vendors say is the next step rather than what the environment can actually support. ZTNA and SASE require clean identity infrastructure as a prerequisite — deploying either without that foundation produces a more complex environment without a proportionate security improvement.

  • Identity infrastructure maturity — ZTNA and SASE require clean identity, enforced MFA, accurate groups and device registration.
  • Application environment — cloud and web applications suit ZTNA and SASE more cleanly than legacy protocols and on-premises dependencies.
  • Remote workforce scale — VPN remains practical for small occasional remote access. ZTNA and SASE fit permanent hybrid work better.
  • Multi-site complexity — branch connectivity, SD-WAN and cloud security requirements determine whether full SASE is justified.
  • Compliance posture — Essential Eight, insurer scrutiny, post-incident remediation and contractor access can accelerate the decision.
  • Operational capability — ZTNA and SASE need ongoing policy ownership, identity hygiene, access reviews and platform operation.
Decision clarity

This is not a question of old versus new.

VPN, ZTNA, Zero Trust and SASE are related, but they answer different questions.

VPNRemote access tunnel

VPN creates an encrypted tunnel that grants network-level access after one authentication event. It is still appropriate for simpler estates: smaller organisations, fewer sites, primarily on-premises applications and limited remote workforce requirements.

VPN also remains practical where identity infrastructure is not yet ready to support ZTNA properly.

ZTNAApplication-level access model

Zero Trust Network Access replaces broad network access with application-level access. Users are granted access only to the specific applications they are authorised for, verified per session by identity, device health and context.

ZTNA is the most direct replacement for user-facing remote access VPN when identity infrastructure is mature enough.

SASEFull network and security architecture

SASE combines SD-WAN with cloud-delivered security services, including ZTNA, Secure Web Gateway, CASB and Firewall as a Service.

It is the broader architecture for distributed, cloud-first organisations that need branch connectivity, remote access, internet security and SaaS visibility under a consistent policy framework.

Zero TrustSecurity principle, not a platform

Zero Trust is not something to buy. It is the principle that access should not be granted implicitly because a user is on a network, has connected through a VPN or has authenticated once.

Access should be verified explicitly, scoped to the application and assessed continuously using identity, device and context.

Architecture models

VPN, ZTNA and SASE assume different things about users, applications and trust.

VPN, ZTNA and SASE represent progressively different assumptions about where users are, where applications live and how trust should be granted.

  • VPN assumes users are occasionally remote and applications are on-premises.
  • ZTNA assumes users are regularly remote or hybrid, applications are cloud-hosted or accessible via an identity broker, and credentials alone cannot be trusted.
  • SASE extends this to the full network and security layer: all user traffic, all branch traffic and all cloud application access under one consistent policy model.
Traditional VPN

VPN creates an encrypted tunnel into the network. After the user authenticates, they usually receive access to a network segment. Access is then controlled by routing, firewall policy and application permissions. The limitation is that the model grants network proximity before application-specific need is verified.

VPN is simple and familiar, but it was designed for occasional remote access — not permanent hybrid work, distributed SaaS use and contractor access.

Zero Trust and ZTNA

Zero Trust is the principle. ZTNA is the access technology that applies it to remote access. ZTNA grants access to specific applications instead of the network. Each session is evaluated using identity, MFA, device posture, role, location and context.

The security value comes from removing implicit network trust. A compromised credential cannot move laterally across the network if the user never receives network-level access in the first place.

SASE

SASE is the architecture that combines network and security functions into a cloud-delivered model. It includes:

  • SD-WAN — branch connectivity and application-aware routing
  • ZTNA — identity-based application access
  • Secure Web Gateway — internet traffic inspection
  • CASB — SaaS visibility and control
  • Firewall as a Service — distributed firewall policy

SASE is usually the direction for organisations where hybrid work, branch connectivity, cloud applications, SaaS governance and identity-led access all need to be managed together.

Comparison

Each model solves a different layer of the problem.

Compare the three models across the same dimensions: what each is, the access model it enforces, where it fits best, its main risk, how it performs and how the transition is approached.

Encrypted remote access tunnel into a network
  • Access model — network-level access after authentication
  • Best fit — smaller environments, on-premises apps, limited remote users, legacy access and site-to-site tunnels
  • Main risk — compromised credentials can enable lateral movement across permitted network segments
  • Performance — can degrade SaaS and Microsoft 365 if traffic is backhauled through a VPN concentrator
  • Transition approach — harden first where replacement is not yet justified
Zero Trust is the principle; ZTNA is application-level access enforcement
  • Access model — application-level access per session, verified by identity, device and context
  • Best fit — hybrid workforce, contractor access, Microsoft 365-heavy environments and VPN replacement
  • Main risk — depends on clean identity, accurate groups, enforced MFA and device posture
  • Performance — better for application-specific access where apps are cloud-hosted or brokered
  • Transition approach — deploy alongside VPN and migrate by user group
Cloud-delivered network and security architecture
  • Access model — consistent access, internet security and branch policy across users, sites and cloud apps
  • Best fit — distributed organisations needing SD-WAN, ZTNA, SWG, CASB and FWaaS under one policy model
  • Main risk — scope overrun if deployed before identity, application and SD-WAN readiness are understood
  • Performance — stronger for distributed SaaS and branch use when cloud-edge inspection is designed properly
  • Transition approach — phase over time: identity, ZTNA, SD-WAN, SWG, CASB and FWaaS as readiness matures
Fit signals

SASE is strongest when users are distributed, applications are cloud-first and the environment is ready for it.

SASE addresses the performance problem VPN creates for cloud applications by routing traffic directly to cloud edges rather than backhauling it through a central concentrator. For organisations where Microsoft 365, Teams and SaaS applications are the primary workload, the performance improvement is architectural rather than incremental.

The security improvement is also architectural. ZTNA removes the implicit network trust that makes compromised VPN credentials so damaging. A credential compromised under ZTNA reaches only the applications that user was explicitly permitted to use and cannot move laterally across the network.

SASE also changes the network operating model, not just the access model. Traffic is inspected at the cloud edge rather than at a central data centre. Branch offices break out directly to the internet under consistent policy rather than backhauling through a concentrator. SaaS pathing improves, user experience becomes more consistent, and policy management shifts from per-device configuration to a centralised policy framework.

VPN remains appropriate where
Simpler estates, hardened properly
  • The organisation is under 100 staff with a single or two-site footprint
  • Applications are primarily on-premises and remote workers are a small minority
  • Identity infrastructure is not yet mature enough to support ZTNA properly
  • Operational simplicity is a genuine constraint and SASE overhead is not justified
  • No specific compliance or security driver is forcing the architecture decision
  • VPN can be hardened with MFA, split tunnelling, tighter segmentation and documented access review
SASE or ZTNA makes sense where
Distributed, cloud-first, identity-ready
  • Hybrid or remote work is permanent and VPN performance is causing friction
  • Microsoft 365, Azure or SaaS platforms are the primary application environment
  • Identity infrastructure is mature, with enforced MFA and clean account hygiene
  • A compliance requirement or security incident has exposed the VPN trust model
  • Third-party or contractor access needs to be scoped to specific applications only
  • Branch connectivity, internet security and cloud application access need a consistent policy framework
Practical scenarios

The scenarios that cover most Australian organisation situations.

The right architecture depends on where the organisation actually sits today. None of these scenarios requires the newest technology. All require an honest assessment of the current environment before selecting a platform. The most common error is deploying ZTNA or SASE as a compliance exercise without resolving the identity infrastructure it depends on. Clean Entra ID, enforced MFA and device registration are prerequisites, not afterthoughts.

01
Smaller, simpler environment — VPN remains the practical answer

This usually applies where the organisation has under 100 staff, one or two sites, VPN working adequately and primarily on-premises applications.

What to do:

  • Harden the existing VPN rather than replace it immediately
  • Enforce MFA on every VPN authentication
  • Implement split tunnelling for cloud traffic where appropriate
  • Tighten network segments accessible via VPN
  • Revisit ZTNA when workforce scale, application mix or complexity grows
02
Microsoft 365-heavy hybrid workforce — ZTNA as VPN replacement

This usually applies where the organisation has 100 to 300 staff, multiple sites, Microsoft 365-heavy applications and hybrid work as standard.

What to do:

  • Deploy ZTNA as a VPN replacement for remote access
  • Confirm prerequisites: clean Entra ID, enforced MFA and device registration
  • Migrate users off VPN progressively, not all at once
  • Keep site-to-site VPN for branch connectivity during transition where required
  • Treat full SASE as the next phase once ZTNA is stable
03
Distributed workforce at scale — full SASE, phased

This usually applies where the organisation has 200 to 600 staff, a distributed workforce, SD-WAN under evaluation and branch connectivity as a current concern.

What to do:

  • Plan full SASE direction over 18 to 24 months
  • Deploy or stabilise SD-WAN first if branch connectivity is a current problem
  • Add ZTNA for remote users
  • Add Secure Web Gateway and CASB as governance requirements mature
  • Use Fortinet Security Fabric where FortiGate is already deployed and the operating model supports it
  • Confirm internal or managed engineering capability to operate the environment after go-live
04
Compliance or security-triggered decision — start with the control, not the platform

This applies at any size where Essential Eight uplift, cyber insurance pressure, post-incident remediation or contractor-access risk is accelerating the decision.

What to do:

  • Identify which control is driving the requirement
  • Map the current access model against that control
  • Resolve identity and MFA gaps before platform selection
  • Scope privileged access and third-party access at application level
  • Avoid using SASE language to satisfy a checkbox while leaving VPN trust risk unchanged
The first step for any organisation moving away from VPN is an identity and application dependency assessment, not a vendor selection.
Failure modes

Most failures come from sequencing, not technology.

01
Deploying ZTNA without resolving identity hygiene first

Stale accounts and MFA exceptions mean ZTNA inherits the same trust problems as VPN. The platform changes. The risk model does not.

Fix: identity remediation runs first, not as a parallel workstream.

02
Treating SASE as a single deployment rather than a phased migration

Attempting all components simultaneously consistently produces scope overrun. SASE is a target architecture. The migration path should be staged.

Fix: sequence identity, ZTNA, SD-WAN, SWG and CASB as readiness matures.

03
Buying SASE licences before assessing on-premises application dependencies

Legacy applications may not be viable ZTNA candidates.

Fix: complete application dependency mapping before platform selection.

04
Setting a hard VPN decommission date before ZTNA access is validated per user group

Hard cutovers without parallel validation consistently produce application access gaps.

Fix: decommission VPN by user group, not by fixed calendar date.

05
Selecting a vendor based on an existing relationship rather than environment fit

This can create a dual-vendor SASE architecture by accident.

Fix: existing stack matters, but platform selection should follow identity, application and operating-model fit.

06
Treating Zero Trust as a product choice

Zero Trust is an access model, identity model and operating discipline.

Fix: stop treating Zero Trust as a SKU. The principle is enforced through policy, identity and operating practice, not purchased through a product labelled Zero Trust.

VPN replacement

ZTNA replaces user-facing VPN. SASE replaces the architecture VPN used to sit inside.

ZTNA is replacing VPN as the remote access model for organisations where identity infrastructure is mature enough to support it. The access layer is shifting from network-level trust granted at the perimeter to application-level access granted per session, per user, per device.

At the broader architectural level, SASE is replacing the combination of MPLS-backhauled WAN, perimeter firewalls and split VPN that many organisations built through the 2000s and 2010s. The architecture that served a world of central data centres and office-bound workers does not serve a world of cloud applications and distributed teams. The same shift is visible at the network layer, where secure SD-WAN is displacing traditional single-carrier WAN designs.

Site-to-site VPN between offices and data centres will remain in many environments for years, particularly where legacy applications cannot be accessed via an identity-aware broker. User-facing remote access VPN is what is being displaced: faster in cloud-first organisations, slower in those with on-premises application dependencies.

Are VPNs becoming obsolete? The decline is real, but uneven:

  • User-facing remote access VPN — being displaced by ZTNA in many Australian environments over 18 to 36 months.
  • Site-to-site VPN between offices — being displaced by SD-WAN overlays within SASE architectures.
  • Site-to-site VPN to legacy data centres — remains appropriate where legacy systems cannot be fronted by ZTNA.
  • VPN in small single-site environments — remains practical where ZTNA complexity is not justified.
  • VPN for specific on-premises application access — may remain in parallel with ZTNA during long migration windows.
Most organisations run VPN and ZTNA in parallel for 12 to 36 months during the transition. Hard cutovers without parallel validation consistently produce application access gaps.
Compliance alignment

SASE and ZTNA can support Essential Eight controls, but they do not automatically deliver compliance.

This comparison often becomes practical when Essential Eight, cyber insurance, insurer questionnaires or post-incident remediation put pressure on remote access controls. SASE and ZTNA can support several Essential Eight controls, but they do not automatically deliver compliance. The outcome depends on identity hygiene, device compliance, group membership accuracy and operational discipline.

Multi-Factor Authentication

VPN with weak or inconsistent MFA is usually the first control gap to surface. ZTNA and SASE improve the model by enforcing MFA and Conditional Access per session, but only if MFA coverage is complete and exceptions are controlled. VPN with SMS MFA does not satisfy higher-maturity expectations where phishing-resistant MFA is required.

Restrict Administrative Privileges

ZTNA's application-level segmentation prevents privileged accounts from having broad network access beyond what they need. Administrative access can be scoped to specific applications and sessions rather than broad VPN access into a network segment.

Application Control

SASE's Secure Web Gateway and CASB provide application-level visibility and enforcement. This covers which SaaS applications are accessible, what data can be uploaded and where AI or browser-based tools sit outside policy.

Patch Applications and Operating Systems

ZTNA's device posture checks can identify unpatched or non-compliant devices before access is granted. This does not replace patch management, but it creates a parallel enforcement point that prevents unmanaged or unhealthy devices from accessing sensitive applications.

Example delivery context

Identity cleanup came before platform selection.

A multi-site branch operations group was Microsoft 365-heavy with FortiGate already at each branch. VPN was causing Microsoft Teams degradation and generating consistent support tickets during peak hours. Before any ZTNA platform was selected, an identity review found 34 stale accounts, MFA disabled for three service accounts and no device registration policy in place. Identity cleanup ran for six weeks before ZTNA deployment began.

ZTNA was piloted with one user group first. VPN remained in place for a legacy ERP application throughout. Full remote-user migration to ZTNA was completed across 14 weeks from pilot start. The VPN concentrator remained active for the legacy application and site-to-site connectivity throughout.

This is the difference between replacing a remote access product and sequencing a secure access transition.

Inlight IT view

The right starting point is an access architecture review before selecting a platform.

The decision between VPN, ZTNA and SASE is not primarily a technology comparison. It is an assessment of where the environment sits today — identity hygiene, application dependencies, site architecture, remote workforce scale, branch connectivity and operational capability. Those inputs determine which platform fits and what work is required before deployment begins.

  • Assess fit before vendor evaluation — organisations that skip the assessment and go directly to vendor evaluation consistently discover, after purchase, that the deployment requires more prerequisite work than the sales process acknowledged. The vendor roadmap describes the ideal-state deployment. It does not describe what the environment needs first.
  • Identity work is the prerequisite, not a parallel project — MFA gaps, stale accounts and on-premises application blockers need to be identified before purchase, not after. The identity remediation work required to make ZTNA function also satisfies several Essential Eight controls at the same time.
  • Most organisations need ZTNA before SASE — the typical pattern is identity remediation, then ZTNA as VPN replacement, then SD-WAN if branch connectivity is a current concern, then Secure Web Gateway and CASB as governance requirements mature. Attempting all components simultaneously consistently produces scope overrun.
  • Run VPN and ZTNA in parallel during transition — parallel operation is not a compromise. It is the correct approach. Most organisations operate both for 12 to 36 months. Hard cutovers without parallel validation consistently produce application access gaps.
  • Match vendor to existing stack — where FortiGate is already deployed, Fortinet Security Fabric is often the operationally cleanest path. Where Microsoft 365 and Entra ID dominate, Microsoft Entra Internet Access and Private Access are the natural progression. Mixing vendors at each SASE layer is viable, but operationally more complex.

Architectural readiness is the benefit. Identity and operating-model discipline is the trade-off.

A useful test

If a compromised password would reach more than the applications that user actually needs, the environment is still running on network-level trust — whatever the remote access product is called.

Test the access model →

Where Inlight IT works alongside the team:

Identity and access readiness review

Entra ID health, MFA coverage, stale accounts, group accuracy and device registration.

VPN and application dependency mapping

Current VPN usage, application access paths, legacy dependencies and user group requirements.

ZTNA design and transition planning

Application-level access, pilot group selection, rollback plan, staged user migration and the VPN coexistence period.

SASE architecture planning

SD-WAN, Secure Web Gateway, CASB, Firewall as a Service, identity policy and the operating model for ongoing policy, identity and access review.

Vendor fit assessment

Microsoft Entra SSE, Fortinet FortiSASE, Fortinet Security Fabric or combined architecture against the current environment.

Essential Eight and cyber insurance alignment

Remote access architecture mapped to evidence, MFA, privileged access and application control requirements, including post-incident remediation drivers.

Common questions

Questions that come up before replacing VPN or planning SASE.

Are SASE and Zero Trust the same thing?
No. Zero Trust is an architectural principle: trust nothing implicitly, verify everything explicitly. SASE is the specific converged architecture that includes ZTNA as one component alongside SD-WAN, Secure Web Gateway, CASB and Firewall as a Service. An organisation can deploy ZTNA without SASE. A fully deployed SASE architecture always includes ZTNA as its access control layer.
How long does a VPN to SASE migration take?
For a 200-staff organisation with multiple sites, a realistic timeline is 18 to 24 months from decision to full migration. Typical phases are identity remediation and device registration over 2 to 3 months, ZTNA deployment alongside existing VPN over 2 to 4 months, progressive user migration off VPN over 3 to 6 months, SD-WAN at branches over 3 to 6 months, and Secure Web Gateway and CASB onboarding over 2 to 4 months. Organisations that compress this timeline consistently encounter application access gaps.
What identity infrastructure is required before deploying ZTNA?
ZTNA requires a functioning identity provider with enforced MFA, clean user account hygiene, no stale or shared accounts, device registration so posture can be assessed at authentication, and group membership that accurately reflects role-based access requirements. Microsoft Entra ID is the most common identity provider in Australian organisation deployments. These requirements should be in place before ZTNA deployment begins.
How does SASE handle on-premises application access?
On-premises application access through ZTNA requires either a connector deployed in the on-premises environment that the ZTNA broker routes traffic through, or an application that can be federated with a modern identity provider. Legacy applications using non-standard protocols may not be viable ZTNA candidates. Identifying on-premises application dependencies before selecting a SASE platform is a prerequisite step.
Does SASE require replacing existing SD-WAN?
Not necessarily. If an organisation has already deployed SD-WAN from a vendor that offers a SASE cloud security stack, such as Fortinet, Cisco Meraki or Palo Alto, the existing SD-WAN investment may become the network layer of the SASE architecture. If the existing SD-WAN vendor has no strong cloud security component, the organisation may retain it and add a third-party SSE layer. That is viable, but more complex to operate.
Does Microsoft have a SASE product?
Yes. Microsoft's Security Service Edge offering is built around Microsoft Entra Internet Access and Entra Private Access, delivered through Global Secure Access. For organisations already using Microsoft 365 and Entra ID, this provides a natural SASE path because identity is already centralised in Entra and integrates with existing Conditional Access policies.
How does ZTNA handle third-party and contractor access?
ZTNA is well suited to contractor access. A contractor authenticated through ZTNA can be granted access to specific applications only, with no network-level visibility into anything else. Agentless ZTNA can support unmanaged contractor devices without requiring software installation, depending on the application and platform. This is a significant security improvement over VPN, which often grants contractors broader network-level trust than they require.
Is VPN obsolete?
User-facing remote access VPN is being displaced by ZTNA in many Australian environments over 18 to 36 months. Site-to-site VPN between offices is being displaced by SD-WAN within SASE architectures. Site-to-site VPN to legacy data centres remains appropriate where legacy systems cannot be fronted by ZTNA. VPN in small single-site environments remains practical where ZTNA complexity is not justified. The decline is real, but uneven.
Does SASE replace SD-WAN?
No. SD-WAN is the network layer within a SASE architecture, handling branch connectivity and application-aware routing. SASE adds the cloud-delivered security stack on top. In a Fortinet deployment, the same FortiGate hardware running SD-WAN can be extended with FortiSASE cloud security without rebuilding the network.
When should we run this comparison properly?
Before a specific event forces the decision. Common triggers include VPN causing daily performance friction, cyber insurance renewal requiring evidence of Zero Trust controls, post-incident remediation, or an Essential Eight assessment flagging remote access gaps. Running the comparison before a trigger appears leaves more architectural options available.
Practical next step

Review access architecture before choosing the platform.

A practical review of identity readiness, VPN dependencies, ZTNA suitability, SASE fit, application access and migration sequence before selecting a platform.

Discuss SASE and Zero Trust