How to Choose and Register a Backup Domain Before You Need It
Backup domain registration means buying and holding a second domain, on a separate registrar account, that you can point your services at if the primary domain is lost. You register it early, configure it ahead of time, and leave it parked until you need it, because the worst moment to discover a problem with a fallback is during an outage.
This is not the same thing as a second DNS provider or a second server. Those protect you against infrastructure failure. A backup domain protects you against losing control of the name itself: an expired card on auto-renew, a hijacked registrar account, a lapsed registration someone else grabs, a trademark complaint that suspends the name, or a billing dispute that locks the account. The name is the thing your users, your certificate authorities and your mail senders all depend on, so treat it as its own single point of failure.
Picking the name for backup domain registration
Choose something that is unambiguous when read aloud and typed from memory. A hyphenated or misspelled name is a phishing vector on a good day and useless in an emergency. Avoid names that differ from your primary by one character, because those are exactly the names attackers register and you will spend your energy on takedowns instead of on recovery.
Prefer a different top level domain from your primary, and check the registry's transfer and dispute rules before you buy. Some ccTLDs impose local presence requirements or slow transfer processes; a fallback you cannot move quickly is not a fallback. Check whether the name has prior history: query it before registering.
dig +short NS backup-example.net
dig +short A backup-example.net
whois backup-example.net | grep -Ei 'creation|registrar|status'
If the name resolves to something already, or the creation date is old and the status shows a hold, walk away. You want a clean, unregistered name. Also confirm it does not collide with an existing trademark in your market, because a backup domain you cannot legally use is worse than none.
Registering it so it survives the failure you are guarding against
The whole point is independence. Register the backup at a different registrar from your primary, and put it on a different account with different credentials and a different recovery email that is not hosted on either domain. If both names sit behind one login, one compromised mailbox takes both. If your recovery email is [email protected], then losing the primary locks you out of the backup too.
Set registration length generously and enable auto-renew, but do not trust auto-renew alone. Keep a card on file that is not the card you use for everything else, and set a calendar reminder to verify the renewal a few weeks before expiry. Registrar auto-renew fails quietly when a card expires, and quiet failure is the exact scenario you are insuring against. Enable registrar lock and two-factor authentication on the account, and store the recovery codes offline.
At the DNS level, you can preconfigure the backup zone so a cutover is a record change rather than a rebuild. Point it at the same infrastructure you would use in a recovery, or at a static holding page. A minimal zone that already answers is far better than an empty one at the moment of need.
; backup-example.net zone
@ 3600 IN SOA ns1.backup-dns.example. hostmaster.backup-example.net. (
1 7200 3600 1209600 3600 )
@ 3600 IN NS ns1.backup-dns.example.
@ 3600 IN NS ns2.backup-dns.example.
@ 3600 IN A 203.0.113.10
www 3600 IN CNAME backup-example.net.
@ 3600 IN MX 10 mail.backup-example.net.
@ 3600 IN TXT "v=spf1 -all"
Note the SPF record. Until the backup is live you want it to authorize nothing, so nobody can send mail as the parked name. When you cut over, replace it with the real policy.
Preparing the cutover before you need it
Two things make a fallback usable. First, the certificates: issue a certificate for the backup name now and keep it renewing, so the site does not serve a warning during a crisis. Second, the app config: make the hostname a variable, not a hardcoded string. If your vhost, your CMS and your mail configuration all name the primary domain in a dozen places, cutover becomes an archaeology project.
server {
listen 443 ssl;
server_name backup-example.net www.backup-example.net;
ssl_certificate /etc/letsencrypt/live/backup-example.net/fullchain.pem;
ssl_certificate_key /etc/letsencrypt/live/backup-example.net/privkey.pem;
return 302 https://primary.example$request_uri;
}
That stub does something useful immediately: it redirects anyone who lands on the backup name to the real site, so the name is never dead. On cutover day you swap the redirect for the real root and proxy configuration, or flip a DNS record. Either way the move is minutes, not days.
Test the fallback on a schedule. Resolve the backup name, confirm the certificate chain, and check that mail sent to it does not bounce unpredictably. A fallback you have never exercised is a guess. Keep the runbook short: which registrar, which login, which record to change, which config file to edit, and who has the credentials. Store it somewhere that does not depend on either domain being reachable.
What to do next
Pick two or three candidate names this week and check their DNS and registration history before committing. Then register the best one at a registrar you do not already use, on a fresh account with an independent recovery mailbox, and set the lock, the two-factor and the auto-renew. Preconfigure the zone and the certificate so the name answers, even if only with a redirect, and write the one-page cutover runbook while everything is calm. The cost of doing this now is small; the cost of doing it during an outage is whatever the outage decides to charge you.
