The CACTOS Vision
Realizing the CACTOS vision means that the variety of workloads supposed to be executed in a Cloud environment can be mapped automatically to the most appropriate resources in the best fitting data centre at a given time and that in case of failures or changing conditions the best matching place is automatically detected and the workload is relocated.
The underpinning idea is based on the assumption that cloud applications can be described and analysed in terms of workload behaviour, potentially split into segments representing different classes of workload and that an optimised placement of the application elements is feasible relying on rich resource descriptions providing the necessary information from server node capabilities over cluster and data centre topology up to environmental data collected by sensors from the facility management system and business data such as actual power costs.
In order to realize this vision, in addition to the existing traces available at the start of the project the consortium will collect further data on the actual use of cloud infrastructures. In parallel, a workload model, based on an analysis of applications as well as on synthetic workloads (with defined behaviour), and an infrastructure model, based on different hardware architecture, further extend the information basis. The top-down and bottom-up models are developed initially in parallel and later integrated into a holistic optimization model. These individual models are validated in the first project cycle using a simulator. Over the project time the models are integrated in several iterations and validated within the selected Cloud Middleware on a physical infrastructure as well as using simulation in order to extend the validation beyond the typical limits in size of prototypical infrastructure.
Contributions to CACTOS
OMI is the coordinator of the project and a work package leader, validating the CACTOS tools. In CACTOS, OMI is operating these tools in an OpenStack Kilo system (~120k€ investment) dedicated for research purposes. Additionally, OMI drives the development of the data collection tools (CactoScale) and the runtime management integration, which is responsible for integrating optimisation actions into a Cloud middleware like OpenStack.
CACTOS Blog Enties
The CACTOS Vision of Context-Aware Cloud Topology Optimization and Simulation
5th International Conference on Cloud Computing
Optimised Cloud data centre operation supported by simulation
eChallenges 2014 Conference Proceedings
Funded as part of the Seventh Framework Programme of the EU with a total budget of 3.45 M€.