Sudden failure in operation of the server, depending on its assignment, is capable to result in unavailability of one or several websites, network printers, FTP directories, etc. Reliability of the server depends both on quality of components, and on observance of service regulations.
Correctly select the processing power of the server depending on load of it. If on it the websites are placed, consider that the they are more popular, the more often contact them. Install software for protection against the DoS-attacks (from English denial pf service). Consider that they will not help with a case of the DDoS-attack (distributed denial of service) which is carried out along with a set of the machines infected with malicious applications. But not viruses are guilty of all such cases. Sometimes the similar effect arises if the reference to the unpopular website is placed on another - popular. Such inadvertent similarity of the DDoS-attack is called Slashdot-effect, in honor of the website Slashdot, placement of links on which often leads to similar situations.
Illiterately configured software of the server can fail at receipt of request with an error. If in URL after a name of a script the question mark, and behind it - the parameters told to a script is located, the script can "take off" when one of such parameters is, for example, too large number. By drawing up a script to it should provide protection against transfer of unforeseen parameters and also saving in logs of the IP addresses of all authors of such requests.
Cases of failures in operation of servers because of overflow of the hard drive therefore it is necessary to select the volume of the last with a stock are frequent. If it suddenly fails, there will be not only a failure, but also data loss - full or partial. To avoid it, periodically make backup copies.
Hangup of the server can happen not only because of the hacker attack or program failure. It can happen also because of poor quality of mains voltage. Therefore it is not necessary to neglect use of such devices as filters and uninterruptible power supply units. The server can hang up also because of dust in its case, especially under the motherboard and also from inflation of condensers which at the same time lose capacity and begin to pass high-frequency pulsations on the processor and other nodes. Also it is not necessary to neglect grounding of the case, not only the electrical safety of the machine, but also stability of its work depends on it. Also hardware sentry timer for automatic restart at hangup will not prevent.
Even completely serviceable and operable server can become unavailable because of failure in operation of the devices which are between it and the user's computer. Therefore before looking for fault in the server, surely make sure that the reason consists not in intermediate switches and routers.