For optimal reading, please switch to desktop mode.

The Manufacturing Technology Centre (MTC) has launched an initiative in partnership with the Liverpool City Region Combined Authority to invest in and develop UK manufacturers and their workforce to successfully build global competitiveness. The Digital Manufacturing Accelerator programme supports companies of all sizes and at all stages of their adoption journey. The MTC do this by providing world-leading engineering experts, alongside access to purpose built digital and physical testbed facilities. MTC expertise and facilities help identify, implement and de-risk the adoption of digital and automation technology solutions to transform businesses and their manufacturing processes. StackHPC engaged with the MTC to deliver a system “design to support” solution based on OpenStack to provide a Digital Factory Environment (DFE). This platform delivers a secure multi-tenant sandpit environment, providing segregated access to resources for multiple industrial partners.

Based on the core infrastructure, the DFE includes an application catalogue to enable self-service deployment of virtual compute platforms with pre-configured software solutions. Likely application workloads include: 3D environments and visualisation, modelling assets in a 3D environment and their interactions with physical properties. For the protection of these industrial partners, the service implements data encryption at rest and in transit.

The rapid creation of software platforms on demand will be essential to delivering expedient results from the platform. A user-friendly portal also plays an important role in helping industrial partners access results quickly. The following use-cases formed consideration of the design:

Simulations and Modelling.

The primary workload for the DFE will be simulations. Such simulations may include physics modelling and discrete event simulations. Simulations are expected to be performed using commercial software packages, which would be selected from the user-friendly application catalogue. Application codes will be deployed in various forms on the OpenStack infrastructure:

• Linux VMs.
• Windows VMs.
• Containerised application deployments, for example using Kubernetes.
• Batch-queued workload platforms.

Machine Learning

Support is provided for AI/ML platforms and workloads, in particular python-based ML runtime environments (such as PyTorch) and Tensorflow. GPU devices will also be available for supporting CUDA and linear algebra workflows for CFD etc.

“Kayobe OpenStack was selected as the infrastructure solution for the DFE through a vigorous down selection process which evaluated the private cloud landscape, taking a capability driven approach to comparing each solution. StackHPC was chosen to deliver our use-case through a series of technical discussions in which their knowledge and expertise reassured our decision-making process by providing evidence of the capabilities we aspired to take advantage of.” – Dr Hannah Edmonds, Chief Engineer, Digital Manufacturing Accelerator @ The Manufacturing Technology Centre Ltd.

StackHPC engaged with the MTC over a long period to:

1. Design a hyper-converged infrastructure solution based on commodity hardware taking into account workflow and workload characteristics; performance requirements; existing compute, storage and networking hardware inventory; scale-related issues; security; integration with existing systems and tooling. This allowed the MTC to separately run procurements for compute, networking and storage hardware.
2. Deploy an OpenStack Kayobe environment in a reliable and consistent method configured and managed as version-controlled source code.
3. Train MTC staff in the day-2 operations of the system with third-level support provided by StackHPC.
4. Maintain the system in such a manner as to ensure the system is kept up to date with the latest versions of OpenStack and Ceph.

“It has been a great pleasure to work with the MTC on this project” says Dr. John Taylor, CEO of StackHPC Ltd. “On-premises clouds, as opposed to public clouds, provide specific virtual organisations a high degree of flexibility, autonomy and data sovereignty to pursue collaborative research and development within a resilient and highly targeted software-defined environment specific to a community. The system built for the MTC will provide the means to strengthen the digital manufacturing ecosystem in Liverpool helping organisations to prototype solutions in a timely manner.”

The MTC is part of the High Value Manufacturing Catapult, supported by Innovate UK, and was established to prove innovative manufacturing processes and technologies in an agile environment in partnership with industry, academia and other institutions. Housing some of the most advanced manufacturing equipment in the world, and with facilities in Coventry and the Liverpool City Region, the MTC provides a high quality environment for the development and demonstration of new technologies on an industrial scale, supporting skills, productivity and growth across the UK manufacturing industry.

StackHPC Limited is a UK-based SME that provides professional services and solutions for high-performance computing (HPC), cloud computing, and scientific and engineering research. The company specialises in designing, building, and supporting HPC and cloud infrastructures for scientific research organisations, government agencies, and private companies. StackHPC's services include consultancy, architecture design, system integration, software development, and support services for HPC and cloud infrastructures. The company also provides custom solutions for scientific data management, machine learning, and artificial intelligence applications. StackHPC has a team of highly experienced HPC and cloud experts who have worked on some of the world's largest and most complex HPC and cloud infrastructures. The company is committed to open source and is actively involved in several open source communities, including OpenStack (through the OpenInfra foundation), Kubernetes, and HPC tools such as Slurm.