The goal of the ATMOSPHERE project is to design and develop an ecosystem comprised of a framework and a platform focused on trustworthy cloud services on top of an intercontinental hybrid and federated resource pool. The framework considers several trustworthiness properties and their measures, while the platform supports the development, build, deployment, measurement and evolution of trustworthy cloud resources, data management services and data processing services. A distributed telemedicine scenario is used to demonstrate the usefulness of the proposed solutions.
A key objective is to design and implement the trustworthiness monitoring and assessment solution for the ATMOSPHERE ecosystem, including a trustworthiness framework, a monitoring platform, measurement instruments, analysis services, and adaptation capabilities. In practice, some specific objectives are:
ATMOSPHERE envisages a model for the a priori and runtime assessment and monitoring of trustworthiness properties allowing: the definition of quality models, to support computing the score of trustworthiness based on the relevant trustworthiness properties and their metrics; the evaluation of different properties of trustworthiness before deployment as well as their dynamic evaluation at run-time based on a monitoring and assessment solution; and the specification of triggering rules that were used to provide trustworthiness guarantees to react to, e.g., security attacks, system failures or performance degradation. Assessing trustworthiness properties requires continuously collecting evidence about the behaviour of the system to compute the relevant trustworthiness scores. Thus a quality model is needed for calculating trustworthiness scores based on the collected data, representing the process followed to perform the data analysis, based on a combination of weights of the attributes that determine the relative importance of each. The trustworthiness model aims at ensuring conformance to the service specification and meeting measurable expectations. Since several conflicting properties may be involved in the analysis, a multi-criteria decision-making (MCDM) based technique is needed.
This deliverable is a result of Task 3.3 - Trustworthiness Measurement and Analysis Services, whose goal is to define procedures and mechanisms to dynamically calculate trustworthiness scores using the data collected by the monitoring platform. In particular, we emphasise the design of quality models for the computation of relevant trustworthiness scores, which will be implemented in the remaining of the project (to be reported in D3.6). Taking into account the trustworthiness properties deemed as more relevant in the context of the project by the different WPs, we develop a set of quality models, detailing the properties / attributes and sub-attributes that compose the corresponding trustworthiness score. Such trustworthiness measurements will not only support the decisions required by the tasks of WP3, but also the ones of the different tasks on trustworthiness assessment and adaptation in WP4, WP5, and WP6.
The deliverable moves forward the state-of-the-art on trustworthiness characterisation in the sense that it defines and discusses a set of trustworthiness properties applicable to multiple scenarios, and shows how quality models can be built and used in practice to compute the relevant scores. These contributions are of key relevance in the context of ATMOSPHERE, but can also be applied in other contexts. In practice, although focused on the project needs, the deliverable is of use for researchers and practitioners working on the design/development of trustworthiness metrics for cloud-based systems, in general.