The second period of the project was focused on the development of the requested features and on the implementation and validation of the scenarios defined, using v1 of the e-Infrastructure as the base for the new works.
The implementation of the EUXDAT e-Infrastructure followed the same approach as in the first period. First of all, in the context of WP2, the requirements and high-level architecture were updated. The previous experience with the v1 of the platform and with the first version of the scenarios, as well as some external feedback collected (i.e. through the hackathons and other interactions with stakeholders) provided the main points for the update of features and architecture. Also, two new scenarios were identified for v3 of the e-Infrastructure, although only one of them was implemented in the end.
Based on the definitions and the updates, v2 and v3 of the EUXDAT e-Infrastructure progressed in the implementation of the main features and the scenarios.
From the end users’ platform perspective (WP3), the implementation included more data connectors (i.e. FTP repository for hyperspectral images, PESSL API, etc.). Also, more libraries were made available for the prototyping environment (Jupyter Notebooks), in line with the needs pointed out by the scenarios and potential utility. The frontend was improved in an important way for accessing easily all the features (including accounting) and for providing a rich GUI for scenarios. The marketplace solution was changed and adapted properly, and both the marketplace and the data catalogue were populated.
The background infrastructure was also further developed (WP4). The tools for enabling easy data movement were set up and integrated in the orchestration mechanism. The applications were defined in such a way they could use HPC and Cloud in an optimal way. The SLA management was set up and the Quality of Service parameters were defined (as well as the SLA templates). Finally, an accounting component was developed, linking it with the orchestrator, the monitoring and the SLA manager.
As for the scenarios, scenarios 1, 2 and 3 were further developed (i.e. HPC usage in the case of scenarios 1 and 3 with new features, usage of the hyperspectral images in scenario 2). Other three scenarios have been developed in v2 and v3: ‘Crop climate risk analysis’, ‘Information support for field use recommendations’ and ‘Crop weather risk monitoring and prediction’. They all have been integrated with the rest of components. The test cases for all components have been defined and executed in order to carry out the TRL qualification.
The dissemination plan has been followed, with participation in social networks, the organization of hackathons, the publication of scientific papers, periodic publication of blogs, etc…
Finally, the exploitation related tasks went on with the market analysis and the selected business models were defined in detail (including also some financial information) together with the services and the targeted stakeholders.
By the end of the project, we consider that the main results of EUXDAT are:
• A platform for end users with a prototyping environment and all the documentation and tools they need to build their own applications for the agriculture domain;
• An orchestration mechanism that not only allows to execute applications in a hybrid environment (mixing HPC and Cloud resources), but also with integrated support for easy data movement;
• A set of ready to use scenarios with innovative algorithms that support an optimal management of crops in the agriculture domain, even in some cases making use of HPC resources.
