User demand and new technologies are driving a drastic increase in cloud infrastructure scale, heterogeneity and complexity. The demand for better energy efficiency has led to a variety of technological options to build servers from different CPU architectures as well as specialised options for highly parallel tasks such as manycore systems or General Purposed GPUs. At the same time, service provisioning has evolved from Web Services and infrastructure virtualisation to Clouds, which is conceptually very similar to the evolution from traditional server hosting to more interactive services (e.g. remote rendering or gaming). Modern offerings go beyond simple services, including full platforms, complex compositions and whole infrastructures. This leads to a significant complexity in mapping the different modules of these solutions on the large variety of available hardware options. Similarly data centres have made significant investments in energy efficient buildings, server racks and facility management technology and understand themselves as Smart Consumers in evolving SmartGrid environments.To cope with the challenge to optimise the mapping of services to a variety of different resources, both hardware and software related (e.g. high bandwidth demands), requires topology aware mapping. This mapping needs to consider placement of the services across geographically distributed centres and demands new intelligent and cross-domain integration of actual and historical usage data.Consequently the key research challenges addressed in CACTOS are as follows:•\tHow to model heterogeneous workloads, infrastructure-architecture models and topologies as well as facility management information and energy supplier information within a coherent information model•\tA scalable and open collection and analysis framework for historical usage data and how to derive from intelligent management strategies how to respond to different observed situations autonomously. This requires integration of research results from the Cloud and Data Centre Management field as well as from Mathematics and Operations Research•\tRealisation of new infrastructure management methods that intregate different aspects into a unified multi-dimensional optimisation model. This includes among others dynamic workload placement, scheduling and migration by continuous optimization across multiple partially orthogonal or correlated criteria (consumer contracts; provider contracts e.g. for energy, network bandwidth, capacity, licenses, ...; energy efficiency and costs)• Development of a simulation framework for conducting costs and risk analysis in order to validate the intelligent Context-Aware Cloud Topology Optimisation strategies for robustness on a large scale beyond the limits of prototypical installations and deployments• The toolkit realised in CACTOS will be validated against three distinct scenarios for business analytics, enterprise applications and technical computing use cases.