How to build a fault-tolerant hosting platform?

Question

High availability

So, you need to make a fault-tolerant hosting platform for your projects.

As done now - bought a dedicated server in hetzner and vds in digitalocean. On the server in hetzner, dnsmanager and vmmanager are running. The 1st dns server is launched in dnsmanager, many virtual machines with their own projects are launched in vmmanager. On the vds in digitalocean, dnsmanager is running as the 2nd dns server.
What is needed is that, when the server crashes, everything continues to work without interference, or at least the most important services, like mail or file storage, work.

Actually, how to organize such work of the system with the least financial costs.

Z.Y. Do not offer to use third-party services (dns-servers, mail, hosting).

Answer 1

As a result, it became clear that real fault tolerance requires what @foxmuldercp described . As a result, duplication of important services (mail, file storage) was made, and backup was set up for everything else.

Answer 2

In general, for good, full fault tolerance is geographically dispersed sites, with a minimum of double duplication of everything on which copies of servers are spinning, with balancing, hot and cold backups and several independent backups far away from data centers in case it gets really hot.
for relatively "simple" services like mail and dns, you need at least two servers according to the standards, but the rest - depending on the requirements and initial configuration - maybe some kind of service, like a storage or subd, you already live in some kind of cloud thread like amazon/azura

Answer 3

Interesting. And you are not confused by the difference in resources on vds for $5 and quite a full-fledged dedicated server)?
On a subject - hire the good administrator. Thousand for 20-50 you will organize it in a normal way.

Answer 4

To be honest, I do not know the provided foreign words, and I do not know any free solutions, but because this is not indicated here, due to the fact that I work in software sales, I know for sure that these tasks are solved by paid VMWare products.

Answer 5

> so that, when the server crashes, everything continues to work without intervention
2 servers with Failover IP and switching at the DC level.