The distributed computing systems widely used in the industry today are too complex for human operators to manually control. This leads to the suboptimal use of resources, recurring failures, and inadequate protection against malicious attacks on system integrity, resulting in a considerable loss of money and a permanent overstrain of operators with all its consequences. According to recent initiatives of market leaders such as IBM, Microsoft, and Sun, the solution to this inherent problem is self-management. They expect that distributed systems will control themselves in the future. The industry hopes for more effective IT management and expects human resources to be freed for other, more important tasks. Here, the authors discuss the current state of the art of self-management and present several open problems whose solution is necessary to make self-management a reality.