Running a redundant DNS provider is expensive as all hell.
While 'expensive' is a relative term, I disagree that it's cost-prohibitive for most firms, as I looked into this specifically (ironically considered using Dyn as our secondary). The challenge isn't coming up with the funds, it's if you happen to use 'intelligent DNS' features; these are proprietary (by nature) and thus they don't translate 1:1 between providers.
In addition to having to bridge the divide yourself, by analyzing the intelligent DNS features and using the API from each provider to simultaneously push changes to both providers, you have to write and maintain automation/tooling that ensures your records are the same (or as close as possible) between the providers. If you don't do this right, you'll get different / less predictable results between the providers, making troubleshooting something of a headache.
Thus in that case the 'cost' in man effort (and risk, given that APIs change and tooling can go wrong) in addition to the monthly fee.
If all you're doing is simple, standard DNS (no intelligent DNS features), it's not as hard, and it's just another monthly cost. Since you typically get charged by queries/month, if you run a popular service you're probably well able to afford the redundancy of a second provider.
Ah so make everything redundant. Double my costs in man hours and in monetary cost. Brilliant!
The sarcasm is curious. It's a business decision. Either your revenue is high enough that the monetary loss from a several-hour intra-day outage is potentially worse than the cost of said redundancy, or you don't care enough to invest in that direction (it's expensive).
Making things redundant is exactly a core piece of what infrastructure engineering is. I guess with the world of VPSes and cloud services, that aspect is being forgotten? And yes, redundancy / uptime costs money!
Your automation should be handling creating/modifying records in both providers. Also, if you're utilizing multiple providers you don't need to pay for 100% of your QPS (or whatever metric is used for billing) on every provider, only 50% for two or 33% for three. You can just pay for overages when you need to send a higher percentage of your traffic to a single provider.
I believe you don't understand DNS. It's probably the most resilient service (granted it's used correctly). There's nothing inherent in the protocol that would prevent them to use multiple DNS providers.
> Running a redundant DNS provider is expensive as all hell.
