High-Performance Computing as a Service: HPC platform on IBM Power systems
Flexible, straightforward, and highly secure: INS offers a Platform as a Service (PaaS) model for Cognitive Computing and Deep Learning, which is unique in Europe. The High-Performance Computing as a Service is based on IBM Power Systems with Nvidia Tesla 100 GPUs, as well as standard software. The HPC platform meets the highest standards for data security and data protection. Operations are conducted from German Tier 3+ data centers.
Our HPC infrastructure and its utilization for maximum project success
The trend in regards to AI applications is moving towards mixed-use infrastructures. According to IDC’s market researchers, 65 percent of AI infrastructures run on-premise, while 43 percent of them are hybrid solutions.
In the US, there already is a special platform for Cognitive Computing and Deep Learning in existence on the “IBM Power on the Nimbix Cloud”. Together with IBM, Tech Data and PlanetAI, INS is bringing the benefits of High-Performance Computing as a Service to German and European companies.
Thanks to the HPC platform ‘Made in Germany’, you can quickly test your theoretical benchmarks in practice without having to utilize or improve/build your own resources. Especially in regards to Proof-of-Concept (POC) applications, this proves to be cost-effective.
On-premise solutions can be combined exceptionally well with our HPC as a Service approach on a later stage during practical operation. You can utilize local resources seamless and scale other resources “on-demand” in the cloud. Thus, benefitting from a new level of flexibility and agility.
Our HPC platform consists of “out-of-the-box” standardized components. This provides you with the opportunity to develop tailor-made AI solutions. At the same time, your data that is located in our Tier 3 and higher certified data centers will be processed safely and in compliance with the law.
HPC Computing on the highest and standardized level
Of course, INS utilizes only the latest and fastest technologies for its High-Performance Computing platform. The core of this solution is the IBM Power8 system. Moreover, four Nvidia Tesla P100 GPUs are connected through an NVLink interface. open source applications, such as Ubuntu and Docker, are preinstalled on the IBM Power system so that you can start on the operating system level right away. Furthermore, we provide you with special software from our partner PlanetAI, which is intended for handwriting recognition, as well as text and speech analysis.
We are happy to assist you in building up your own solution, which is individually tailored to your needs. Our AI experts support you during all project phases up to the full operation of the systems including backup, archiving and disaster recovery. Benefit from our close cooperation with Tech Data’s AI team, which provides added value especially in regards to design and implementation.
Our PaaS model in detail
The test data and production data is provided by you, whether in physical form, on a storage device, or from within the cloud. Moreover, you create and supply the application container. We deliver a suitable management tool for the Docker container to you. als Tenant.
The connection to the system is established via VPN, SFTP or HTTPS. In addition, the entire database is stored on a suitable NFS storage. Moreover, temporary memory can be added at any time.
Operation, availability and backup are defined via Service Level Agreements (SLA).
Strong reasons for HPC as a Service
Our High-Performance Computing as a Service is the correct choice for you, if…..
… you want to test your AI application as part of a Proof of Concept (POC).
… you only require resources temporarily.
… you want to increase the flexibility and scalability of your AI infrastructure.
… you want to rely on a competent partners regarding security and compliance.
… you are looking for external support in regards to backup or operation.
Have we piqued your interest? Then we are looking forward to hearing from you!