Cloud & DevOps
Domain and DNS Setup That Won't Bite You Later
Updated December 16, 2015By the CalliArc team
Key takeaway
Own the domain in the company's name with an account nobody personally controls, enable auto-renew and registrar lock, and put the renewal reminder on a monitored mailbox. Most domain disasters are administrative — an expired registration or an unreachable former employee — not technical.
Domains and DNS are set up once in an afternoon and then forgotten until something breaks spectacularly. A few decisions at the start prevent almost all of the common failures.
Ownership first
- Register in the company's legal name, with a company email address on a mailbox more than one person monitors — not a developer's personal account or an agency's.
- Enable auto-renew, keep a valid payment method on file, and check it annually. Expired domains are recoverable but expensive and slow.
- Turn on registrar lock to prevent unauthorised transfers, and enable multi-factor authentication on the registrar account.
- Keep the WHOIS contact reachable; it's how you'll be told about a dispute or a problem.
- Register the obvious variants and common misspellings if the brand matters, and redirect them to the canonical domain.
Records you'll actually use
- A and AAAA point a name at addresses; CNAME aliases one name to another, and cannot exist at the apex alongside other records — use your provider's alias record type for the root domain.
- MX for mail delivery, which must survive any web hosting change; separating DNS management from hosting is wise for exactly this reason.
- TXT records for domain verification and for email authentication — SPF, DKIM, and DMARC — which are what keep your mail out of spam folders.
- CAA to restrict which certificate authorities may issue certificates for your domain.
TTLs and change management
- Lower the TTL a day or two before a planned migration so the change propagates quickly, then raise it again afterwards.
- Remember that a low TTL only helps if you set it before the change, not during it.
- Keep an exported copy of the full zone. Rebuilding DNS from memory during an incident is a bad afternoon.
- Treat DNS changes with the same care as a deployment: reviewed, recorded, and reversible.
The failures worth monitoring
- Domain and certificate expiry, with alerts weeks in advance to a team address.
- Dangling records pointing at decommissioned services, which can be claimed by someone else — a real subdomain takeover risk.
- Nameserver reachability, ideally with providers in more than one network if the business depends on availability.
- Email authentication records after any change of mail or marketing provider, which is when they most often break silently.