I currently host around 30 zones with ClouDNS and have generally been happy with it, but DNS changes can take several minutes to reach all authoritative locations. This is especially inconvenient for ACME DNS-01 validation, where I usually need to wait at least three minutes and sometimes closer to five.
I tested this by creating unique TXT records through the APIs and checking them locally and from multiple regions. Across ten rounds, Bunny DNS had a median authoritative propagation time of about 3.6 seconds, with a worst case of 5.5 seconds and successful results from all 120 checks. ClouDNS took about 27 seconds at the median and 35.5 seconds at worst from my local checks. Even after local servers showed the new record, several ClouDNS anycast locations still returned NXDOMAIN, with only 72 of 120 worldwide checks succeeding.
Bunny has become more appealing since I last evaluated it, with a proper web interface, DNSSEC, and API support. My main concern is reliability: ClouDNS advertises roughly 65 DNS PoPs, while Bunny has more than 36 dedicated DNS PoPs, and Bunny has experienced some notable platform outages. However, Bunny also appears more transparent when incidents occur.
Is anyone using Bunny DNS in production? Have other ClouDNS customers noticed similarly slow or inconsistent propagation? I am also interested in how people handle DNS-01 validation and whether using both providers or delegating an ACME challenge subdomain makes sense.
3 Answers
I have used ClouDNS for about eight years and Bunny DNS for roughly six months. I use Bunny for less active domains and ClouDNS for the rest, partly because both APIs make centralized automation straightforward. I have not personally seen propagation delays as severe as yours, even while doing a lot of ACME work. Many people simply use a larger general-purpose DNS provider, but your need for separately managed subzones can make that less attractive.
I have run Bunny since 2020 and have not experienced any significant DNS or CDN outages. It has been stable enough for my production workloads, and I am happy with the service. A provider's PoP count is only one factor; actual operational consistency and how clearly incidents are reported matter just as much.
You could make the ACME automation retry with increasing delays instead of relying on one fixed wait. Another option is to delegate an acme-challenge subdomain for each service to a faster DNS provider while leaving the main zones where they are. That gives you quick DNS-01 updates without moving every record.
The separately maintained subzone is exactly why I cannot use some providers without paying for an enterprise feature. Are you using ClouDNS for DNS-01 validation, or only Bunny for that?