This document discusses challenges in deploying distributed applications on cloud infrastructure and proposes an autonomic framework called Scarce to address them. It notes that applications' components may be unevenly distributed across virtual machines with varying performance. Scarce uses autonomous agents that monitor components and make placement decisions. It employs an economic model where servers charge components rent based on resource usage, and components aim to maximize their balance of earnings and costs through actions like replication and migration. The framework also propagates service level agreements from parent to child components and automatically provisions resources to ensure performance guarantees are met under varying load. Evaluation results demonstrate its ability to adapt to changing loads and failures while maintaining scalability.