Norwegian Artificial Intelligence Cloud (NAIC)
The aim of NAIC is to establish the most powerful AI infrastructure in Norway with a competent staff to operate and support it. At the same time the usability aspect, support for user onboarding and project specific assistant will be given a very high priority.
The project implementation is divided into three phases: during the first phase all AI/ML hardware resources that are currently in use and are planned to be made available for the consortium will be cataloged and a uniform load monitoring system will be implemented. The aim of this is to find out the immediate needs and to evaluate whether a load balancing solution could be used to distribute the load, thus making the existing resources usage optimal before procuring any new ones. An emerging technology review will also be conducted in parallel to find out the best technology to acquire to fill any gaps. In addition recruitment of new staff and further building their competence on operating the infrastructure and designing courses for the users will be initiated. During the second phase, the technology solutions decided during phase one will be implemented and staff and user training and user support will be established. During the last and third phase, a central high performance infrastructure will be set up according to the requirements discovered during the first two phases. After the end of the funding period the infrastructure will be transferred to SIGMA2 and further funding opportunities will be actively investigated for the continuation of the project.
NAIC will be a distributed AI computing and storage infrastructure encompassed by experienced expert staff to provide training, documentation, helpdesk, a knowledge hub and operations.
Stage one: Catalogue all AI/ML hardware resources currently in use and are planned to be made available for consortium. Implement load monitoring and investigate cross institute user access procedures. Feasibility report on using the resources at disposal as a pool to support overall load to load spikes. Competence building for staff. For implementation of the proposed solution and train-the-trainer for planned user training. Expand training already provided by the partners and introduce new course. Emerging technology technology assessment work to discover potential future procurement. Gap filling infrastructure procurement.
Stage two: Implement a federated access model or single-sign-in solution enabling Norwegian researchers to get access to any of the partner institutes that will provide such resources as in-kind. If federated access is impossible due to location-specific technicalities those sites will still enable access through some more cumbersome methods. Start the user training and support desk. NAIC will utilise the partners´ already established course, training and information channels, CMS systems and not at least through the NAIL and NORA networks. Part 2 includes education and training of NAIC staff (train the trainers). Design the FAIR data storage solution, implement it and arrange user training
Stage tree: Main procurement and operations
Depending on the experience and competence acquired during the first two years, provide access to final infrastructure as IaaS (infrastructure as a service) providing virtual machine provisioning, an IaaS using a queue system to give fair access or a combination of the two methods. Plan further funding and transition phase for handling operations over to Sigma2 “Driftsorganisasjonen”