Like many others in the IT sector, high-performance computing (HPC) facilities are beginning to recognize the advantages offered by the public cloud. Greater agility, lower costs, access to new capabilities – there’s a lot to like from the cloud. Interestingly, “new revenue source” hasn’t often been one of the advantages mentioned by HPC decision-makers. But maybe it should be.

Established over 230 years ago, MINES ParisTech is a leading French graduate school that houses 18 research centers in five fields: earth sciences and environment; energy and process engineering; mechanical and materials engineering; mathematics and systems; and economics, management, and society. It has more than 1,280 students at locations in Paris, Évry, Fontainebleau, and Sophia Antipolis.

MINES ParisTech and software development firm Transvalor originally partnered to provide HPC services to internal customers as a way to drive research forward and promote the development of future industry leaders. Then the partners discovered that the appetite for HPC services extended beyond the campus boundaries.

School IT decision-makers recognized the powerful opportunity to help accelerate research and development across a much broader horizon of organizations and industries – while at the same time creating a revenue source based on the basic HPC investments already made by the university.

MINES ParisTech could become an HPC cloud services provider to both internal and external customers. But transforming such an idea into reality would require some help.

First, management and scheduling of HPC resources in a cloud model is different from how these crucial tasks are often accomplished in more traditional HPC environments. The self-service nature of the cloud means that users without in-depth infrastructure knowledge and skills will expect to easily submit work and get results, as well as provision and schedule resources transparently. To be a successful cloud services provider to both internal and external customers alike, MINES ParisTech must implement a simplified, intuitive user interface that could hide underlying complexity while orchestrating constantly evolving customer requests and HPC resources availability.

And what happens when significant customer requirements overlap and HPC resources run low? Certain priorities and service-level-agreements must be met. These scenarios of spiking demand are actually one of the most common drivers of interest in the cloud model in the first place. Rather than build HPC infrastructure to meet any demand – which would prove prohibitively expensive – data centers are turning to public cloud resources during peak demand times. Hybrid cloud solutions allow organizations to keep capital investments at reasonable levels while enabling them to address spikes in customer demand – and pay for the extra resources only when used.

But hybrid cloud solutions bring a range of challenges and complexities all their own – from compatibility issues to security concerns – not to mention the management nightmares of manually starting up resources at a cloud provider’s site at a moment’s notice, then quickly shutting it all down again when waves of peak demand pass.

And a common cloud issue arises when very large HPC data sets must be moved or copied across the network to support external customer jobs. During these data transfers, the cloud resources meter is running, but no real customer work is getting done.

MINES ParisTech IT architects understood these and other challenges of becoming successful enterprise-class HPC cloud services providers. They turned to a trusted IT advisor, CARRI Systems, who knew the university’s HPC environment well. CARRI Systems brought in IBM to provide the technology and expertise needed to help the school and software firm navigate the complex HPC cloud.

The solution providers recommended IBM Spectrum LSF Suites. This HPC software-defined infrastructure (SDI) technology from IBM brought proven, mature, leading-edge solutions to the key challenges faced by the French HPC cloud services providers:

• IBM Spectrum LSF Suites provides powerful workload scheduling and management functions that enable improved productivity across the school’s HPC environment.
• The solution hides complexity, with an interface that enables users who are experts in fields other than technology – engineers or physicians, for example – to access data quickly and easily to conduct research and analysis. It also allows users to interact with HPC resources via a tightly integrated client for Microsoft Windows environments and via mobile clients for Google Android and Apple iOS platforms.
• IBM Spectrum LSF Suites can dynamically provision external cloud resources until peak loads have passed, enabling HPC capacity to grow – and shrink – in response to demand, so customers only pay for what they use.
• And with IBM Spectrum LSF Suites, data is transferred before jobs are dispatched, enabling users to avoid data migrations that occupy compute resources. And when multiple jobs need the same information, data it is transferred from its source location only one time and then cached.

Finally, implementing the SDI-based IBM Spectrum LSF solution was easy and cost-effective. Romain Klein, Cloud Architect at Transvalor, notes: “IBM Spectrum LSF Suite for Workgroups is designed to be set up rapidly, so with help from CARRI Systems we were able to get it installed and running within a very short time.”

The HPC cloud services solution from CARRI Systems and IBM Spectrum Computing is helping MINES ParisTech attract new students to the school by enhancing the school’s reputation as a leader in research, and aiding the partnership attract new external customers by meeting demands for performance, scalability, and low cost. 

As Elie Hachem, Head of the Computing and Fluids Research Group at MINES ParisTech, states: “Our IBM solutions play an essential role in helping us to fulfil our goal of advancing industry and innovation within France, and eventually Europe.”

Learn more about how IBM Spectrum Computing helped a university in France expand its HPC capabilities, its cloud customer base, and its revenue streams.