Security / ZTNA / VPN Replacement / Procurement / TCO

Zero Trust Network Access (ZTNA) Buying Guide 2026

When to replace VPN with ZTNA, how to shortlist Prisma Access and Zscaler ZPA, identity prerequisites, license meters, and a dual-vendor RFQ checklist.

Quick Take

Quick Take: ZTNA replaces broad VPN tunnels with identity-aware access to named apps. Choose it when remote users and private apps dominate and your IdP is ready for MFA and posture. Dual-RFQ Prisma, Zscaler, Fortinet, and Cisco on equal seats over a 3-year TCO window; keep a scoped 90-day VPN bridge while app cohorts migrate. If about 90% of staff never leave campus, delay a full estate cutover.

ZTNA vs VPN: The Architecture Fork

Quick Answer: VPN extends the network to the user. ZTNA brokers access to named applications after identity and device posture checks. The fork is trust scope: broad tunnel versus least-privilege app access. Buy ZTNA when remote users should never land on a flat internal segment by default.

Buying ZTNA seats without cleaning identity groups and retiring overlapping VPN licenses creates double spend and leftover lateral paths. Equalize users, apps, and posture before award.
4
Platform paths
3-yr
TCO window
App-first
Migration model
12+
Guides interlinked

Legacy remote access assumed the user was trustworthy once authenticated. A VPN tunnel then exposed routes, printers, and lateral paths that attackers love. ZTNA flips that model: authenticate the user and device, then open only the application path required for the job.

That does not mean VPN disappears overnight. Jump hosts, OT consoles, and some thick clients still need broader adjacency. The 2026 pattern is ZTNA-first for standard knowledge workers, with a tightly scoped VPN exception list that shrinks each quarter.

Write the fork into the brief. If a vendor quotes ZTNA seats but still designs a full-tunnel default, you bought new branding for old risk. Acceptance tests should prove app-level access, not only login success.

When ZTNA Wins for Mid Estates

Quick Answer: ZTNA wins when remote and hybrid users dominate, private apps sit behind connectors, and identity is clean enough for MFA and groups. It shrinks VPN sprawl and improves auditability for security teams. Delay a full cut if most users are on-site and private apps are few.

Cloud-first workforces feel the win fastest. Users move between home, hotel, and branch, and they need SAP, Git, or admin consoles without a permanent LAN presence. ZTNA maps that reality better than hairpin VPN concentrators.

Compliance and M&A also push ZTNA. Segmenting contractors onto named apps is easier than giving them a network route. When two companies merge, ZTNA connectors can expose selected systems without flattening both networks.

If about 90% of staff never leave campus and private apps are local-only, a full ZTNA platform may be overbuy. Start with contractor and VIP remote pools, then expand over 6–12 months. Partial coverage still beats a rushed all-or-nothing cut.

Platform Shortlist: Prisma, Zscaler, Fortinet, Cisco

Quick Answer: Shortlist Prisma Access, Zscaler ZPA, Fortinet Unified SASE, and Cisco ZTNA-capable stacks for procurement in 2026. Pick by existing firewall adjacency, PoP coverage, and identity fit, not brand alone. Dual-vendor RFQ at equal users and apps before you lock a five-year platform.

Prisma Access is natural when Palo Alto already owns NGFW and Panorama. Policy language and ops habits transfer. Zscaler ZPA is natural when you want a broker-first cloud SSE plane with strong private-app isolation.

Fortinet Unified SASE appeals when FortiGate already sits at branches and you want one commercial motion. Cisco paths matter when Duo, Umbrella, or Secure Access already define identity and DNS security. None of these is universally cheaper until meters match.

Demand PoP latency samples for your user regions, connector sizing guidance, and a clear map of ZTNA versus SWG entitlements. A low seat price that forces a second product for internet security is not a low TCO.

Identity, MFA, and Device Posture

Quick Answer: ZTNA quality tracks identity quality closely in practice. Require MFA, group-based app grants, and device posture signals before production. Weak IdP hygiene recreates VPN risk inside a modern console. Budget identity remediation as part of the ZTNA program, not as an afterthought.

Connectors and PoPs are the easy part. Dirty Active Directory groups, shared accounts, and missing MFA break least privilege. Clean the identity model first: named users, role groups, and joiner-mover-leaver automation.

Device posture separates managed from unmanaged access. Decide which apps allow BYOD with browser isolation or limited roles, and which require corporate posture. Document that matrix in the RFQ so vendors quote the right controls.

Service accounts and API integrations need a parallel plan. If automation bots still rely on network-local trusts, ZTNA for humans alone leaves a hole. Include non-human access in the discovery workshop.

Deployment Modes and Connector Placement

Quick Answer: Most designs place lightweight connectors near private apps, then steer users to the nearest cloud PoP. Choose cloud-delivered ZTNA for speed, or hybrid when sovereign and latency rules demand local control. Always pilot one business unit carefully before any estate-wide cutover.

Connector placement decides performance. Put connectors close to the application tier, with enough capacity for peak concurrent sessions. Under-sizing connectors creates the same complaints users had about slow VPN concentrators.

DNS and split tunneling design still matter. Users should reach SaaS directly when policy allows, while private apps stay on the ZTNA path. Hairpinning all traffic through a single region destroys the latency benefit of global PoPs.

Run a production-like pilot: real apps, real MFA, real posture failures. Measure login time, app open time, and failure handling. A lab demo that never fails posture checks teaches nothing about support load.

Equalize License Meters Before Award

Quick Answer: ZTNA is usually billed per user per month; VPN is often per concentrator or concurrent session. Compare at equal headcount, apps, MFA, and posture features over three years. Strip overlapping VPN, SSE, and ZTNA seats for the same people before you call either path cheaper.

Sales decks hide meter tricks. One quote includes ZTNA only. Another bundles SWG, CASB, and DLP. A third keeps perpetual VPN plus add-on ZTNA. Force a feature file: private app access, posture, MFA integration, logging, and HA.

Count active remote users, not total employees, unless everyone truly needs access. Then model growth. Per-user cloud fees look fine at five hundred seats and painful at five thousand if VPN was already paid.

Ask for an entitlement export of what you already own. Many estates discover unused SSE seats that can cover early ZTNA phases. Competitive pricing helps after the meter is honest, not before.

Migration Scope: Apps First, Then Users

Quick Answer: Migrate by application cohort, not by big-bang user cutover. Publish connectors, map IdP groups, move one cohort, keep VPN fallback, then expand. Legacy protocols and thick clients are the critical path. Plan rollback as hard as the formal go-live checklist.

Discovery beats configuration. Inventory every private app, protocol, and owner. Mark HTTP-friendly apps as wave one. Mark SMB, legacy thick clients, and jump hosts as wave two with exceptions.

User communication matters. Explain why the old full-tunnel VPN is shrinking and how to request an exception. Support queues spike when people discover missing apps on Monday morning without a clear path.

Keep a measured VPN bridge for ninety days after each wave. If ZTNA connectors fail, users still work. Remove the bridge only after ticket volume and success metrics stabilize.

RFQ Checklist for ZTNA Procurement

Quick Answer: List remote user counts, private app inventory, IdP and MFA stack, posture rules, PoP regions, concurrent peaks, three-year TCO meters, and VPN sunset criteria. Require dual-vendor responses on equal seats and features. Reject quotes that cannot show app-level access proofs in pilots.

Copy into the brief:

  • Remote and hybrid user counts, plus contractor population.
  • Private app list with protocols and data sensitivity.
  • IdP, MFA, and device management stack in use today.
  • Required PoP regions and maximum login latency targets.
  • Posture rules for managed versus BYOD access.
  • Three-year seat math and overlapping VPN or SSE entitlements.
  • Acceptance tests: named app access, denied lateral routes, MFA failure handling.

Award only after two vendors clear the same tests. That rule ends slideware comparisons.

Common Mistakes in a 2026 ZTNA Buy

The first mistake is treating ZTNA as a VPN SKU rename. The second is skipping identity cleanup. The third is awarding on seat price without PoP latency and connector capacity proofs. Fix scope and meters, then compare platforms.

Another mistake is promising one hundred percent VPN retirement on day one. Legacy protocols need an exception list with owners and end dates. Without that, shadow VPN tunnels return within weeks.

Finally, do not ignore logging and SOC handoff. ZTNA only improves security if deny and allow events land in the same investigation workflow your team already trusts.

Frequently Asked Questions

Quick Answer: ZTNA is the default remote-access path for 2026 refreshes: access brokers per application on identity and posture, so lateral movement shrinks and least privilege becomes auditable. Keep a scoped VPN for legacy and thick-client protocols, insist on a capable IdP with MFA before buying seats, and strip overlapping VPN licenses once cohorts migrate.

For most remote-access refreshes, yes as the primary path. ZTNA grants application access by identity and posture instead of a flat network tunnel. Many estates keep a thin VPN fallback for legacy protocols while new users land on ZTNA first.

VPN usually places a user on a broad network segment. ZTNA brokers access to named applications after identity and device checks. That shrinks lateral movement risk and makes least-privilege easier to audit in an RFQ acceptance test.

Both deliver strong ZTNA. Prisma Access fits Palo Alto-heavy estates and Panorama adjacency. Zscaler ZPA fits cloud-first SSE estates that want a pure broker model. Dual-RFQ both on equal users, apps, and posture rules before award.

Usually yes. ZTNA covers private app access for users. Data-center east-west, internet breakout, and IPS decrypt still need NGFW or SSE SWG. Most designs keep firewalls inside and push remote private access to ZTNA.

A modern IdP with MFA, group claims, and preferably device posture signals. Without clean identity, ZTNA becomes another VPN with extra steps. Budget IdP readiness and conditional access before you buy seats.

Yes. Fortinet Unified SASE and Cisco with Duo or Umbrella paths can cover ZTNA for many mid estates. Trade-offs are PoP coverage and feature depth versus Zscaler or Prisma. Ask for a same-meter quote next to a specialist SSE option.

Buying per-user ZTNA while keeping full VPN licenses and overlapping SSE seats for the same population. Equalize active users, apps, and posture features, then strip shelfware before the platform award.

Pilots often run two to four weeks for a critical app set. Broad rollouts run one to three months depending on legacy protocols and connector placement. Plan rollback and a temporary VPN bridge as hard as the cutover.

Sometimes with connectors and protocol support, sometimes not. Catalog every non-HTTP app early. If a thick client needs full network adjacency, keep a scoped VPN or redesign the app path before you promise one hundred percent ZTNA.

Yes. Share user count, app inventory, IdP stack, and current VPN spend for a BOM risk comparison.

Key Takeaway

Key Takeaway: Shortlist by adjacency: Prisma Access for Panorama estates, Zscaler ZPA for broker-first SSE, Fortinet Unified SASE for FortiGate branches, Cisco for Duo/Umbrella estates. Equalize per-user meters on active remote headcount over three years, strip overlapping VPN and SSE seats, and migrate app cohorts behind a ninety-day VPN bridge. BENQUAN supplies Fortinet and Palo Alto platforms with competitive pricing.

Ready to RFQ ZTNA Against Your VPN Estate?

Share remote user counts, private app inventory, IdP stack, PoP regions, and current VPN or SSE entitlements. We will build a matched, dual-vendor BOM risk comparison across Prisma, Zscaler, Fortinet, and Cisco-capable paths.

Request a Quote
Contents