Advice · Integrate · Manage

From a real access need to production and ongoing governance.

Identify the right access path, choose Advice, Integrate or Manage, and see how responsibilities, controls, acceptance tests and evidence become part of a production-ready Tailscale environment.

Choose by your current state

What is the next decision that needs to be completed?

Choose the stage based on whether the target model is still open, an approved solution needs production delivery or a live environment lacks ongoing ownership.

Three-question service selector

What stage is your access path in right now?

The selector uses only the answers below. Nothing is stored or sent to a server.

1 Are the target model and first use case approved?
2 Is the agreed access path built, tested and documented?
3 Does the live tailnet need an external ongoing owner?

Advice · decide

When direction or scope is still open

Outcome
A decision-ready target architecture, access model and implementation sequence.
Delivery
A scoped fixed-price project with up to three workshops.
We need from you
An owner, the key use cases and current-state information.
Boundary
No production configuration or ongoing operations.

Integrate · deliver

When the target model is approved

Outcome
A production access model that is tested and documented.
Delivery
A scoped or phased implementation project.
We need from you
An owner and agreed access to identity, device and network environments.
Boundary
No 24/7 operations or full management of other platforms.

Manage · govern

When the tailnet needs an ongoing owner

Outcome
Policies, devices, changes and documentation governed on an agreed cadence.
Delivery
A monthly service, separate onboarding and a six-month minimum term.
We need from you
Administrative roles, approvers, a change process and service boundary.
Boundary
Finnish business days 08:00–17:00; no SOC, MDR or incident response.

A good starting point

A service fits when an access path needs a technical decision or operational owner.

  • A public management port, broad VPN permission or vendor access needs to be restricted.
  • Tailscale is in pilot or production, but policy, tests or responsibilities are unclear.
  • The customer has a named owner and can participate in decisions.

Not the right service

We do not promise outsourcing of all security or every platform.

  • The requirement is a 24/7 SOC, MDR, incident response or general help desk.
  • The goal is an automatic compliance opinion or legal interpretation.
  • Production changes would need to proceed without an owner, recorded starting state or approved rollback.

The change visible to the customer

Less unnecessary visibility. Clearer access to the right resource.

The value of access comes from combining technical connectivity, business approval and ongoing ownership into one manageable system.

Reduced attack surface

The management service does not need to be exposed through a public ingress port.

The selected SSH, RDP or other administrative connection can be implemented over an encrypted Tailscale connection.

Restricted visibility

The user connects to a resource, not the entire network by default.

Grants policy defines the source, destination and allowed protocol and port.

Governed lifecycle

Access has an approver, owner and removal process.

User and device changes are tied to the customer's identity and endpoint processes.

Verifiable control

Allowed and denied behaviour is tested before acceptance.

The access matrix, policy, test results and change path remain available to the customer.

Before ↔ after

See the difference: from network access to a justified access path.

Switch views with the buttons. The target model does more than encrypt traffic: it limits who can connect, from which device, to which resource and with which network capability.

Before · network visibility as permission

One VPN permission exposes more than the task requires.

The connection is encrypted, but the decision remains network-wide. Destination scope, device trust and access ownership may depend on separate processes.

  1. 1SourceVPN user
  2. 2ConnectionVPN gateway
  3. 3Visibilitymultiple subnets
  • Decision: the user receives network connectivity.
  • Test: the VPN connects.
  • Potentially unclear: whether every visible destination is required.

After · identity, device and resource scope

An approved user and device receive only the named connection.

The business decision becomes a testable Grants rule. The allowed connection works; named wrong users, devices, destinations and ports are denied.

  1. 1Identitygroup:ops
  2. 2Device conditionmanaged = true
  3. 3Grantsrc → dst
  4. 4Resourceprod-ssh:22
  • Decision: the operations group needs SSH access to one server.
  • Test: the allowed path works and the wrong port is denied.
  • Evidence: approval, policy diff, test result and rollback version.

Select your situation

Five common access paths, one shared decision model.

Each case states who or what connects, from where, to what and under which condition. Acceptance always includes named denied connections.

Administrator → cloud server

Remove the public SSH or RDP management interface.

The administrator connection is tied to an approved identity, an agreed device condition and a named server. Other destinations are not allowed by the same Grants policy.

Current state
Public port, jump host or broad VPN connection
Target
Approved administrator and device → server A → tcp:22 or tcp:3389
Acceptance
Agreed administration works; the wrong user, device and destination are denied

Vendor → restricted customer environment

Give a partner access to the resource, not the entire network.

Vendor access receives a business owner, technical scope and removal process. Any time limit is integrated with the customer's identity and approval process.

Current state
Shared account, permanent VPN access or manual port opening
Target
Named partner group → agreed device condition → one service
Acceptance
The intended partner has access; removal, ownership and denied destinations are tested

Employee → internal application or database

Replace broad network visibility with a resource-specific connection.

An approved user group acts as the policy source. A device condition can be included when reliable posture data is available from the customer's endpoint management.

Current state
A VPN user can see multiple subnets and services
Target
Approved group and work device → application or database → required port
Acceptance
The connection required for the task works; other destinations are not allowed by policy

Site or legacy device → cloud or data centre

Connect a subnet without installing the client on every device.

A subnet router represents resources behind it. Device-specific identity and posture do not extend to them in the same way as native Tailscale nodes, so the responsibility boundary is documented.

Current state
Site-to-site VPN or difficult-to-maintain routing
Target
Managed subnet router → named subnets, sources and ports
Acceptance
Routes, availability design, connection type and allowed sources are tested

CI/CD or workload → production resource

Give automation a machine identity, not a person's account.

A tagged workload receives only the network connectivity required for its task. Auth-key or OAuth lifecycle, tagOwners and secret handling are designed as part of the implementation.

Current state
Static IP allow-listing, shared password or personal account
Target
Tagged workload → named production service → restricted protocol and port
Acceptance
The automation path works; key rotation and removal are documented

Choose a decision, implementation or ongoing ownership

Three services, three explicit outcomes.

Choose the phase based on the decision or ownership that is missing now.

Delivery model

Decision, production and day‑2 governance form one continuum.

  1. 1

    Advice makes the target explicit.

    The use case, owner, dependencies and acceptable target state are decided before production changes.

  2. 2

    Integrate implements and tests the boundaries.

    The agreed policy enters production, allowed and denied paths are tested, and rollback is documented.

  3. 3

    Manage keeps the model current.

    Changes, reviews, exceptions and reporting form an ongoing control cycle.

Explore the delivery package

See what a decision, implementation evidence and ongoing-governance material can look like.

Every view is an illustrative, anonymised example—not material from a real customer or a customer reference.

Illustrative · anonymised example

DF / DEC-01Draft for decision

Decision memo: administrator access to a production server

Need
Administrator SSH without public ingress
Choice
Native node + Grants + managed-device condition
Owner
Infrastructure service owner
Acceptance
Allowed path works; another group and port are denied

The decision names the need, owner and acceptable boundary before technical policy changes.

Illustrative · anonymised example

DF / PATH-01Approved model

Access-path record

  1. Sourcegroup:ops
  2. Conditionmanaged device
  3. Destinationtag:prod-ssh
  4. Capabilitytcp:22
Approver
Resource business owner
Removal
Remove group membership + repeat test

Source, condition, destination, network capability, owner and removal method form one reviewable unit.

Illustrative · anonymised example

DF / POLICY-17Change for review

Grants policy diff

- "dst": ["tag:production:*"]
+ "src": ["group:ops"]
+ "dst": ["tag:prod-ssh"]
+ "ip":  ["tcp:22"]

The diff shows exactly which permission was removed and which scoped permission replaced it. This excerpt is conceptual, not deployable policy.

Illustrative · anonymised example

DF / TEST-01Acceptance evidence

Allowed and denied tests

✓ ALLOWEDops + managed device → prod-ssh:22Expected: connects · Result: connects
✓ DENIEDops + managed device → prod-ssh:443Expected: denied · Result: denied
✓ DENIEDanother group → prod-ssh:22Expected: denied · Result: denied

The positive test proves usability. Named negative tests show that the boundary also works for the wrong source or capability.

Illustrative · anonymised example

DF / RACI-01Responsibility boundary

Concise responsibility matrix

TaskCustomerDefense First
Business permissionApprovesDocuments
Grants changeAuthorisesImplements
Identity and deviceOwnsMaps condition
Threat responseOwnsOutside service

Technical delivery does not transfer the customer's identity, business, data or incident-response responsibilities.

Illustrative · anonymised example

DF / RB-01Rollback guide

Rollback: policy change

  1. 1
    Trigger

    The allowed production path fails acceptance testing.

  2. 2
    Restore

    Publish the last approved policy version from version control.

  3. 3
    Verify

    Repeat allowed and named denied tests; record the result in the change trail.

When the trigger, version to restore, required access, operator and post-rollback test are agreed before the change.

Illustrative · anonymised example

DF / MONTH-04Manage review

Monthly report

3approved changes
0open critical exceptions
2stale devices removed
  • Decided: vendor access ends on the agreed date.
  • Next: review the device group with the accountable owner.
  • Boundary: SIEM alerts and incident response remain with the customer.

The report separates completed changes, open decisions and customer-owned controls so monthly-service coverage stays unambiguous.

Controls and evidence chain

A technical rule is tied to an approved need and a completed test.

  1. 1

    Decide

    The resource owner approves the purpose, scope and duration of access.

  2. 2

    Enforce and test

    The decision becomes policy and is checked with both an allowed connection and named denied connections.

  3. 3

    Retain and review

    The rationale, approver, diff, test result and rollback version are retained together.

Responsibilities

Platform capability, Defense First deliverables and customer responsibility are separated.

Key access-control responsibilities
LayerDefense FirstCustomer
Identity and deviceMaps groups, tags and posture conditions into the access model.Owns users, MFA, MDM/EDR data and approvals.
Connection and policyModels, implements and tests the agreed Tailscale boundary.Approves business access and controls other network routes.
Logs and responseConfigures agreed log sources, export and review.Owns retention, SIEM, alerting and threat response.
Application and dataDocuments the network-access boundary.Owns application roles, data and action-level authorisation.

Evidence boundaries

A technical access control supports compliance but does not create it on its own.

Defense First delivers the agreed technical material and states its boundaries. Overall risk, legal or standard interpretation, and identity, endpoint, application, data and threat-response controls remain separate assessments.

Frequently asked questions

Key commercial and technical boundaries before work starts.

Is Advice always required before Integrate?

Not necessarily from Defense First. Integrate does require an approved target model, scoped use cases and known dependencies. Missing decision work is offered as Advice.

Can you improve an existing tailnet?

Yes. Existing policy, roles, tags, routes, devices and ownership are assessed before changes. Working connections are not changed without an approved plan.

Is the current VPN or access route replaced all at once?

Not necessarily. A new access path can be built alongside the old route, tested and accepted before the old route is removed. The change and rollback method depend on the current architecture.

Who owns the tailnet and configuration after delivery?

The customer owns the tailnet, approved policy, access matrix, tests and operational documentation. Administrative roles, approvers and rollback permissions are agreed during delivery.

Is Manage a 24/7 monitoring service?

No. Coverage is provided from 08:00 to 17:00 Europe/Helsinki on Finnish business days. Manage does not include 24/7 on-call coverage, SOC/MDR or incident response. Exact response targets and boundaries are stated on the Manage service page.

Next step

Scope the right first delivery.

The discussion identifies your current phase and selects Advice, Integrate or Manage.