I'm not claiming it's impossible, just saying its commonly frowned upon and industry wide regarded as bad practice.
If your node goes down, so does your DNS. You won't be able to failover your domains cos guess what? That's also down.
Mail? Down cos your node is tripping.
Then there's the security issue. It's pretty easy to tear down a webserver. When that webserver goes down due to excessive load, so does your DNS.
A single network issue, which isn't uncommon, will bring down your entire cluster.
Then there's the speed issue. One server, can't use geodns, anycast, load balancing just to name some options.
In short your created not one but probably at least 10 SPOF. We cannot all afford an infrastructure like DAN but you can create at least some degree of 'high availability' which is quite affordable. If you charge what you charge, you should at least provide your clients with a secure and stable infrastructure.
I don't see why people would pay at all for what you're offering. Plenty of good, reliable DNS services out there for next to nothing.
Unless you need your DNS to do something custom, people are better off just using a service like
@Name Trader asked about. Heck, even your registrar's DNS will outperform whatever your 'setup' is.
That you didn't encounter any issues yet is either because you're lucky or because it went unnoticed. 1300 doesn't mean much. It all depends on the amount of traffic, stack and lander you use. DNS doesn't stress a server that much, it's usually the webserver.
And yes, there's a very tiny chance all your servers go down. Will never be 100%. What you can do is plan ahead for issues. Using hot spares, spinning up a new server the moment another one crashes. Easydns has this nifty feature where they update your DNS servers to working ones when they go down.