Cloud computing is a paradigm for enabling remote, on-demand access to a set of configurable computing resources as a service. The pay-per-use model enables service providers to offer their services to customers in different Quality-of-Service (QoS) levels. Service Level Agreement (SLA) is a negotiated agreement between a service provider and a customer where QoS parameters specify the quality level of service that the service provider have to guarantee. However, due to the dynamic nature of the Cloud and its instability, some SLA violations can occurred and the service providers can be charged for penalties. In this paper, we aim at addressing the Cloud instability to better control SLA management (in particular SLA violations) and indirectly the Cloud elasticity. We propose CSLA, a new SLA language directly integrating some features dealing with QoS uncertainty and Cloud fluctuation. In our evaluation, we present a novel profit model for service provider and new algorithms (for admission control and scheduling) to meet SLA requirements (e.g. prevent SLA violations) while tackling scalability and dynamic issues.