D3.2 – WP 3 Scientific Report and Prototype Description – Y2

This deliverable presents the Scientific Report and Prototype Description of the work carried out during the second year (Y2) of the project, within WP3, to develop the Data-Driven Infrastructure Management (DDIM) capability of the BigDataStack environment.

The document presents the details of the solution implemented during Y2 as well as the experimental results obtained from the research conducted to bring intelligence to the DDIM, that is, on the component ranking and enacting the best application deployment configurations (ADS Ranking and Deploy) and the component making re-deployment decisions (Dynamic Orchestrator).

In particular, the specific machine learning (ML) algorithms used to bring data-driven decisions to the infrastructure self-adaption process are explained, as well as the research conducted to validate them.

The rest of services playing a more conventional role in the infrastructure management are also presented: from the management of the infrastructure resources (Cluster Management and Information-Driven Networking) to the monitoring of those resources as well as the performance of the application and data services making use of them (Triple Monitoring and QoS Evaluation).