An important characteristic of cloud infrastructures is scalability on demand. A scalability service monitors performance load metrics and decides to scale up or down, by provision or revoke of cloud resources. This could guarantee Quality of Service (QoS) and enforce Service Level Objectives (SLOs). The approach of this paper shows that with additional imprecise information (e.g. expected daytime performance) the up and down scale mechanism of such an infrastructure can be improved and SLA violation can be avoided.