Lamersky question

We now have dedicated hosting on Agave. On the night of 13 to 14 the server ceased to be. First of all, of course, I thought that the payment for the domain name might be over, I checked - there is another six months. I didn’t remember by heart ip at that moment (I remember :) now, but I found something (for dacha, beer, basma in the rain ... terrible gprs connection) - ip- no answer, I thought that the server payment might be over. I found it difficult to enter the client part - everything is normal, I wrote a question to the round-the-clock support service, the answer did not immediately wait. I called and found out that it was hot in Moscow (in Novosibirsk on the street at that time it was +5) and the server in the data center overheated and you need to leave a request for a reboot (to the question how often this answer was - last summer). I did not wait for a response from the server - I fell asleep. In the morning I read the letter: “There were problems with temperature in the data center. The server has shut down due to overheating. Now enabled, available on the network. " And while not a single word about "we apologize ...".
In this regard, there is a question that I think will not only be interesting to me:
“Dear community, tell me how the server structure should be built for the site to be in constant access?”