[ Business value ]

Plan a supplier exit before delegating DNS

Make supplier exit a condition of the initial design. Identify the authority being granted, the evidence needed to operate independently and the responsibilities that must survive termination.

By the EmailAuth research team · 6 min read

[ Key points ]

  • NS delegation, record aliases and reporting services create different dependencies.
  • Preserve usable exports and provenance without placing private keys in the handover packet.
  • Sequence a transition from observed conditions rather than a universal propagation deadline.
  • Keep exit charges, assistance and workload assumptions separate until contractually verified.
On this page

Plan the way out before the way in

An authentication supplier can become a dependency long before anyone signs an exit notice. A delegated DNS subtree, vendor-hosted policy target, reporting mailbox or signing arrangement may continue influencing mail while the management dashboard is unavailable. Before granting control, identify what must keep working if the relationship ends unexpectedly. This guide proposes an exit assessment; it does not approve delegation or describe verified EmailAuth migration features.

Start with the business service rather than the vendor account. Record the domains used for employee mail, transaction messages, marketing and non-sending assets, then identify the systems that publish or consume their authentication data. A useful exit plan says who can restore each dependency, where the replacement configuration comes from and how the team will establish that legitimate mail still authenticates.

Draw the authority boundary accurately

Keep registration, authoritative DNS hosting and application administration distinct. Under RFC 1034, a zone cut delegates authority for a subtree. An NS delegation therefore has a different scope from a CNAME that makes a particular name depend on another name. Neither is merely permission to view a dashboard. Document the exact owner names, record types, targets, parent control and affected descendants before evaluating either arrangement.

Include dependencies that cross the apparent boundary. A selector alias can depend on a vendor's DNS and signing service. An SPF include can make authentication depend on records maintained elsewhere. A reporting destination may remain operational without controlling policy at all. For each connection, distinguish the supplier's ability to change an answer from the customer's ability to replace or withdraw the connection. If DNSSEC is deployed, commission a separate validated chain-transition plan rather than improvising DS changes during termination.

Make the exit package usable without the supplier

Specify the exports needed to operate independently: domain and sender inventories, published configuration, approval history, unresolved exceptions, report provenance and retention metadata. Ask for formats, field definitions, timestamps, completeness checks and a way to verify an export before cancellation. A spreadsheet of dashboard totals may be inadequate if the new operator needs original report identifiers and observation periods to avoid counting the same activity twice.

Current aggregate-reporting guidance addresses duplicate and overlapping reports. Preserve enough provenance to reconcile old and new collectors. Confirm who may receive exported information, how long it may be retained and who authorizes deletion. Do not put private signing keys or reusable account credentials in a general handover document. Where signing remains with another platform, record the custodian and approved transition process instead.

Design a transition that can be observed

Prepare the replacement before removing the old dependency, subject to the actual services and change approvals. A replacement DNS answer must express the intended policy or key, not simply recreate a convenient screen label. RFC 9989 makes authenticated identifier alignment central to DMARC; copying a record does not prove that an application's messages use an aligned identity.

Record the relevant authoritative and resolver observations, earlier cache lifetimes, scheduled changes and planned checkpoints. A newly lowered TTL does not shorten data already cached under an older value. Negative caching can also affect recently introduced names; RFC 2308 defines its own handling. Avoid a universal propagation deadline. State what evidence will justify advancing, pausing or recovering, including the possibility that the old supplier is no longer available.

A fictional example uses billing.example and a vendor-managed DKIM selector. The proposed exit requires the sending platform's owner to identify a replacement signing arrangement, the DNS owner to review the exact record change and the mail team to verify representative messages afterwards. No records, accounts or messages in this example were queried. Keeping the old public selector for an agreed transition may be appropriate, but a compromised key demands a different recovery decision.

Price responsibilities rather than a vague migration

Build a requirement-and-evidence matrix with one row per dependency. For each row, record the customer owner, supplier obligation, evidence requested, acceptance condition and unresolved issue. Include normal termination, supplier insolvency, loss of administrator access and emergency suspension. A requirement without a responsible party is an assumption, not an exit control.

Use transparent workload estimates rather than a fabricated migration fee. As a fictional planning calculation, reviewing twelve domains at twenty minutes each is four hours before testing, coordination, correction and sign-off. That arithmetic says nothing about actual EmailAuth effort or price. Show the assumed domain count, complexity, labor rate if approved, overlap subscriptions and any contractually verified export charge separately. Include a range for unresolved cases and identify which uncertainty could dominate the schedule.

NIST CSF 2.0 places supplier requirements in agreements and includes activities after a relationship ends in GV.SC-05 and GV.SC-10. Apply those outcomes through negotiated responsibilities; the framework does not promise a vendor's assistance, export entitlement or cancellation terms. Commercial and legal owners must verify the applicable agreement.

Set a grant and release decision

Before granting delegation, require evidence that the customer retains appropriate parent-zone control, that another qualified operator can use the exit package and that the proposed recovery path does not depend entirely on the departing supplier. Approve only the exact authority needed. Keep access revocation and data deletion sequenced so that evidence is preserved and continuity is checked before destructive steps.

The catalogue requirement remains open: A requirement-and-evidence matrix and transparent workload or cost assumptions. The map below is a proposed structure, not a completed customer matrix. Publication as a demonstrated supplier-exit case requires actual authorized inventory, contractual confirmation, reviewed assumptions and acceptance evidence. A documentary plan can expose the right questions now without claiming that a real migration has succeeded.

[ Takeaway ]

Approve the dependency only after its owner, replacement path, evidence export and recovery decision are understood.

[ By role ]

What this means for you

Choose your role for practical next steps.

DNS and mail operations

Map the recoverable dependency

A replacement must preserve the intended authentication behavior, not merely replicate a control-panel label.

Next steps

  1. Record each delegated subtree, alias, policy dependency and signing custodian.
  2. Define observable checkpoints and a recovery path that works if the supplier is unavailable.
  3. Verify representative legitimate mail after separately authorized changes.

Procurement and legal

Put exit responsibilities in the agreement

Commercial expectations must be distinguishable from rights and obligations actually agreed.

Next steps

  1. Verify export format, availability, fees, assistance and termination notices.
  2. Review retention, deletion and permitted recipients of exported data.
  3. Separate normal termination from emergency suspension or supplier failure.

Security owner

Approve only necessary authority

A convenient service design can create excessive control or a fragile recovery path.

Next steps

  1. Review the exact authority and data access requested.
  2. Confirm customer control and an independent recovery owner.
  3. Require acceptance evidence before revoking access or deleting transition records.

[ FAQs ]

Frequently asked questions

Is an NS delegation equivalent to adding one CNAME?

No. An NS delegation establishes authority for a subtree, while a CNAME makes a particular name an alias. Map the exact names and dependent services before deciding what control is being granted.

Which access should the customer retain?

Identify who can control the parent zone, registration and relevant sending platforms. Verify the actual accounts and recovery process; this guide does not establish current access or authorize changing it.

Can a dashboard export be enough for migration?

Only if it contains the fields the next operator needs and can be interpreted independently. Preserve report identifiers, periods and provenance where aggregate data must be reconciled across collectors.

Should private DKIM keys be included in the exit spreadsheet?

No. Record the key custodian and approved transition procedure instead. The public key and selector inventory are different from private signing material.

Does lowering the TTL make an immediate cutover safe?

No. Previously cached data may retain its earlier lifetime, and new names can be affected by negative caching. Use an observed transition plan and application checks, not a fixed waiting period.

Does a supplier promise to help because an exit plan mentions it?

No. Export rights, assistance, charges and deadlines need confirmation in the applicable agreement. NIST supplier-risk guidance supports planning these responsibilities but supplies no vendor entitlement.

How should uncertain exit effort be estimated?

Show the domain count, task frequency, complexity and assumed time separately. Add testing, exceptions and coordination, using ranges for uncertainty. The article’s twelve-domain calculation is fictional planning arithmetic.

What is still needed before this becomes an evidenced exit case?

A completed requirement-and-evidence matrix, transparent workload or cost assumptions, actual authority inventory and approved acceptance evidence. The proposed map does not close those commissioning requirements.

[ Sources ]

Sources

  1. NIST Cybersecurity Framework 2.0
  2. RFC 1034: Domain Names Concepts and Facilities
  3. RFC 2308: Negative Caching of DNS Queries
  4. RFC 6376: DomainKeys Identified Mail Signatures
  5. RFC 9989: Domain-Based Message Authentication Reporting and Conformance
  6. RFC 9990: DMARC Aggregate Reporting

[ Glossary ]

Glossary

Zone cut
A boundary at which authority for a DNS subtree is delegated to another zone.
CNAME dependency
Reliance on DNS resolution of an alias target rather than delegation of an entire subtree.
Exit package
The proposed set of usable inventories, configurations, approvals and evidence needed by the next operator.
Report provenance
Information identifying where an aggregate report came from and which period and observations it represents.
Transition acceptance
An agreed evidence-based decision that the replacement dependency meets its operational requirements.

Enterprise provenance

Why top financial institutions and governments retired their email security dashboards.

Traditional email authentication platforms sell you a dashboard and ask your engineers to do the labour. In high-stakes environments, manual oversight means delayed enforcement, configuration drift and unmonitored shadow domains.

EmailAuth is the agentic email provenance platform. Our agents enforce p=reject, keep third-party senders aligned and move your domains to full RFC 9989 compliance, the new DMARC standard,, with human-in-the-loop auditability.

Governments Large enterprises Tier-1 financial institutions

SaaS · Dedicated tenancy · Air-gapped on-prem · BYOK encryption

Knowledge Base