An Integrated Optimisation Framework for Reliable Capacity Provisioning for Distributed Clouds – A Conceptual Reference Model

There is an exponential proliferation of interconnected smart objects, both static and mobile, serving different use cases and vertical markets. Against this backdrop, service providers and their users have increasingly heterogeneous Quality of Service (QoS) and Quality of Experience (QoE) requirements that must be maintained to avoid economic penalties. The complexity of managing the ICT operations to meet the unpredictable, fluctuating and mobile demand of these smart objects is putting a strain on existing cloud computing and network infrastructure that was not designed for this purpose. It is no longer realistic for ICT teams to cost-effectively foresee and manage manually the reliable capacity provisioning of all possible configurations, component interactions, and end-user operations from the cloud to the edge on a detailed level. New concepts and technologies are required for the automated orchestration of applications and the end-to-end management of networking infrastructure and workload placement from the cloud to the edge.

This white paper describes a model-based framework for automated optimisation whereby near real-time and future application placement and infrastructure optimisation from the cloud to the edge is achieved, balancing application provider and infrastructure provider interests, and optimising the placement of resources and functionality to improve performance and reduce costs.

The Architectural Components of the RECAP framework describes the important aspects of performance analysis and monitoring, application optimisation, infrastructure optimisation, and orchestration identification.

The Modelling Framework of the RECAP framework describes five major model types need for reliable capacity provisioning for distributed clouds –Infrastructure Models, Workload Distribution Models, Workload Translation Models, User Models and Quality of Service models. These models have a key role in augmenting workload arrival patterns to predict how load propagates through distributed applications, and how the component workloads impact the infrastructure resources running them.

The Optimisation Processes describe how the modelling framework and architectural components interact to optimise applications and infrastructure on distributed clouds.

This conceptual reference architecture is intended as a quick reference for describing, discussing, and developing components or systems using a common framework of reference. Additional reference materials is available through the RECAP website (https://recap-project.eu).