While the simplified example we started with (the unlucky administrator at a party—remember him?) might not frighten many, there are more nightmare scenarios available than we'd want to think about. Various services can die without a sign until it's too late, and a single memory leak can bring the system down easily.

We'll try to explore the available options for making sure such situations are detected as early as possible in order to, say, help our administrator deal with disk space problems during the working day and not find out that an important service has died because of a database hiccup just as they go through the door.