RESuME high-performance computing (HPC) and storage solution

Tender description

At this time, the National Manufacturing Institute Scotland (NMIS) is exploring the potential to create a suite of intelligent and connected systems in order to illustrate and test the possibilities from a streamlined supply chain network within Scotland. This testbed is called RESuME which will comprise of hardware and software solutions aimed at integrating manufacturing equipment and supporting systems across NMIS specialist facilities (AFRC, LMC and Digital Factory) under the ISA-95 model. NMIS is currently looking to purchase on-premise high-performance computing (HPC) and storage solution. The HPC solution will comprise of 2 elements: - On-premise hardware solution, consisting of multiple CPUs, GPUs, and with flash storage for the OS partition and disk storage for local data. The processing capability should be sufficient for an enterprise level HPC (approximate order of 10 petaFLOPS in AI applications), GPUs to be CUDA-enabled and multi-instance GPU capability is desirable. - Software solution installed on hardware solution, capable of operating a container-orchestration environment such as Kubernetes or similar, supporting initially 50 users (with expectation to grow to 250+ over 3 years), capable of hosting 3rd party software and API’s. The storage solution will operate as a Data Lake and will comprise of: - One main storage node within the AFRC, consisting of hot and cold storage. - Two satellite storage nodes within 2 other NMIS facilities (locations TBC) - Software solution installed across all nodes to allow management and access, providing global name space, access to many different data/transfer protocols, tiering storage, lifecycle management, data recovery, encryption and meta data management.