PRIMERGY CDI
Composable Disaggregated Infrastructure
Huge amount of information is generated from our life. To utilize such information for our business or our life, various data processing is required to computer systems. The new generation technology to address data processing diversification is called 'Composable Disaggregated Infrastructure (CDI)'.
PRIMERGY CDI is constituted of high-quality, reliable PC server PRIMERGY, GPU by NVIDIA®/storage which are disaggregated into hardware resource pool shared via high-speed fabric and exclusive software to manage them. PRIMERGY CDI enables efficient operations and maximizes resource utilization by flexibly deploying and releasing resources according to customer workloads.
"PRIMERGY CDI V1.1 powered by Red Hat OpenShift Container Platform", which is a new additional lineup, includes several new features, including autoscaling, which monitors workload changes and automatically adds and releases resources per logical server.
Enhancement of PRIMERGY CDI will continue.
* PRIMERGY CDI (V1 and V1.1) is available only in APAC and Japan regions.
Our Commitment
Automatic Optimization Technology
Dynamically assemble logical servers with embedded high-performance GPUs as needed. Reduce power consumption of overall system.
Use of Standard Technology
Standard technology for container infrastructure will be used. Effective use of existing application resources will become possible.
Large-scale Parallel Technology
Technology developed in supercomputers which realizes workload separation, identification of fault location and maintainability is implemented.
Product Configuration
From a minimum configuration of 1 PCIe device to a maximum configuration of up to 30 PCIe devices, PRIMERGY CDI offers a variety of configurations based on customer requirements.
Refer to the PRIMERGY CDI System Configuration Manual and related documentation for details on the PRIMERGY CDI V1.0, PRIMERGY CDI V1.1, and PRIMERGY CDI V1.1 powered by Red Hat OpenShift Container Platform, including product configurations, the number of compute servers that can be combined, and the number of devices that can be installed.
In addition, for detailed information on configurations, options, peripheral equipment, etc. , please refer to the PRIMERGY CDI System Configurator via the link below.
Applicable Scenes
IaaS environments/private clouds
Even in IaaS environments and private clouds where multiple users are accessing at the same time, you can reduce wasted resource allocation throughout the system by ensuring user security while automatically scaling resources as each user's workload changes with the autoscale feature on a per-logical server basis.
* The autoscale function for each logical server is a "PRIMERGY CDI V1.1 powered by Red Hat OpenShift Container Platform" function.
BIG DATA ANALYSIS
For example, gather multiple GPUs in 1 compute server while AI training, allocate 1 GPU to each compute server while AI inference. In this way, configuration by component is possible. This enables efficiency even in systems that integrate multiple uses.
Download
System Configuration
Whitepaper
 About the Performance of PRIMERGY CDI |
| PRIMERGY CDI GPU Scalability |
| PRIMERGY CDI Performance Comparison of Generative AI Inference Benchmarks |
Manual
Please download* manuals such as 'System Configuration Manual' and user's guide from Product Support page.
*Before downloading, please go to Product Support page and press 'Select a new Product>>' button, select 'Product search', enter 'PRIMERGY CDI', then press 'Continue>>' button.
