Data-intensive workloads have seen a revolution after a series of technological changes that can be described in two main waves: A first wave has brought accelerators and non-volatile memories into the field, with the introduction of GPUs, FPGAs and NVMe cards in some of the most demanding computing challenges; A second wave (disaggregated computing) refers to a computer organisation model in which resources are not exposed as pre-configured computers inspired in the Von Neumann architecture, but instead, pools of resources are packaged together (memory-dense nodes, storage-dense nodes, accelerator-dense nodes), interconnected with very fast network fabrics, and supervised by a global system manager that can dynamically interconnect those resources to assemble general purpose nodes on-demand.
In this context, Software Defined Infrastructures (SDI) provide roots to dramatically increase the efficiency in the management of computing infrastructures. SDI platforms feature mechanisms to get accelerators and other resources to be dynamically attached and detached from physical nodes with just a software operation, removing the constraints of getting hardware components confined to the servers where they are physically connected.
Today, several domains of technology have already taken advantage of the first wave of advances in computing technologies. The second wave is still under R&D phase, and therefore we are in front of a greenfield for developing new solutions. Through the development of HiOMICS, we consolidated the development of a SDI orchestration platform to that can take advantage of SDI flexibility, providing dramatic improvements in terms of infrastructure cost-efficiency compared to the most advanced accelerator-based platforms currently existing. The technology can be applied to several sectors, from genomics to 5G/IoT, and has been successfully transferred for exploitation to a spin-off company created by the hosting institution and the PI.
