Skip to main content

Core product 2

Post-Quantum VPN Migration Programs

Convert an approved migration roadmap into controlled vendor-specific changes, staged migration waves, verification evidence and operational handover.

Or schedule a call

Prerequisite

Migration starts from a verified assessment

PQVPN does not begin a migration programme without a verified inventory and a per-tunnel capability position. Not as a commercial gate — the assessment can be one you already have, or one your own team produced.

The reason is operational. A migration planned against an unverified inventory fails in the change window, on a device that turned out to be running a different release than the record said, in front of an audience. Every hour spent confirming the estate beforehand removes several from the recovery.

Delivery models

You choose how much PQVPN touches

The deliverables are identical across all three. Only the hands differ.

Customer-led

PQVPN produces the artifacts, runbooks and test plans. Your engineers execute every change. PQVPN reviews the evidence afterwards if you want a second pair of eyes.

Credentials: None required.

Joint delivery

Work splits by platform or by wave. PQVPN takes the platforms where specialist knowledge saves the most time; your team takes the rest and gains the knowledge as it goes.

Credentials: Scoped, time-bounded, or none — your choice per platform.

PQVPN-led

PQVPN delivers the programme end to end under your change control. Your team retains approval on every window and receives full handover at the end.

Credentials: Scoped and time-bounded, or executed through a runner you host.

Workflow

How a migration wave runs

  1. Confirm desired state

    Agree the target mechanism per tunnel, and what evidence will prove it.

  2. Validate prerequisites

    Confirm release, licence, peer capability and fragmentation behaviour before anything is scheduled.

  3. Generate implementation artifacts

    Produce the configuration, the diff, the test plan and the rollback plan together, as one reviewable unit.

  4. Peer review

    A second engineer reviews the intended end state, not only the syntax. Nothing reaches staging without this.

  5. Stage changes

    Load into the estate's own change system, in its own format, under its own approval.

  6. Canary or pilot

    One or two low-criticality tunnels first, on the platform combination that covers most of the estate.

  7. Migrate by wave

    Ordered by risk and dependency, with HA members always in separate changes.

  8. Verify

    Read the negotiated SA from both peers. Configuration is intent; the SA is the result.

  9. Roll back when required

    A written, tested rollback exists before every window opens. Stopping is a normal outcome.

  10. Hand over

    Updated inventory, evidence set, runbooks as executed, and the deferred register with owners intact.

Implementation methods

The right mechanism depends on the platform

There is no single tool that manages every enterprise network platform, and claiming otherwise produces migration plans that stall on contact with the estate. The method is chosen per platform, from what actually works there.

MethodWhen it applies
TerraformCloud VPN resources and vendor platforms with a maintained, capable provider
AnsibleFleets of similar devices with a reliable connection plugin
REST APIVendor platforms with a documented configuration API
NETCONF / RESTCONFNetwork operating systems with a stable model for the relevant configuration
Vendor management platformEstates already governed centrally, where local changes would be overwritten
Vendor-native configurationPlatforms whose own tooling is the only reliable path
Reviewed CLI plansDevices with no viable programmatic path — generated, diffed and human-reviewed, then applied
Linux configuration managementstrongSwan and other software gateways, including overlay terminators

Credentials and access

Access models, in order of preference

Customer executes
PQVPN generates the artifacts; your engineers apply them. No PQVPN access to any device.
Customer-hosted runner
Artifacts execute from a runner inside your environment, under your credentials and your audit trail.
Temporary credentials
Scoped to the devices in the wave, valid for the change window, revoked at the end of it.
Vault or KMS integration
Credentials brokered through your existing secret management rather than handed over.
Vendor manager integration
Changes applied through the management platform that already governs the estate.

Unsupported devices

What happens to the ones that cannot move

Some devices will not reach the target on any supported release. That is a finding with options, not a dead end, and each option is costed against what the tunnel actually carries.

OptionNote
Firmware or software upgradeCheapest real fix where the model can run a supporting release
Licence changeWhere capability is gated commercially rather than technically
Hardware refreshAlign with a refresh already in the budget wherever possible
Overlay gatewayTerminate a migrated tunnel on a gateway you control, in front of the legacy device
Compensating controlApplication-layer protection, with its own post-quantum position established
Reduce what the tunnel carriesMove the sensitive flow rather than migrating the tunnel
Defer and trackNamed owner, stated reason, review date. A legitimate outcome when it is explicit
Partner coordinationWhere the constraint is a counterparty rather than a device

Each option is examined in detail in options for legacy VPN gateways.

Verification

Proving what was negotiated

Verification reads the key exchange method actually negotiated on the established security association, from both peers. It does not re-read the configuration that was pushed.

The distinction matters because the common failure is silent. A proposal that still offers a classical-only option will negotiate that option, the tunnel will establish, monitoring will stay green, and the change will be recorded as complete. Nothing alerts.

Evidence is also captured after a failover and after a rekey, because both are separate code paths from establishment.

Risk controls

What sits around every change

  • Preflight checks against the actual device, not the inventory record
  • Configuration diffs reviewed by a second engineer
  • Change windows agreed inside your existing governance
  • Written, tested rollback before every window
  • Canary tunnel ahead of every new platform combination
  • Traffic validation from the far side, not only tunnel status
  • Routing validation, including re-established dynamic sessions
  • HA validation with a real failover test where policy permits
  • Evidence capture from both peers, stored against the change record

Deliverables

What the programme produces

  • Vendor-specific implementation artifacts
  • Change plans in your own format
  • Test plans per platform
  • Rollback plans, written before the window
  • Validation evidence from both peers
  • Updated topology diagrams
  • Runbooks as executed, not as drafted
  • Handover workshop with your engineers

Questions

Migration FAQ

Discuss a migration program

Tell us where the estate is now — whether an assessment exists, and who you want executing the changes. The scoping form takes a few minutes and returns a preliminary complexity band.

Or schedule a call