The rapidly growing use of analytical forecasting systems, big data analysis, AI and deep learning requires higher efficiency, significant cost savings and significant reductions in the environmental impact of data centers. High Performance Computing (HPC) is an important tool to balance IT budget, productivity and efficiency. For this, the computing power per rack space must be increased considerably.
Traditional data centers typically only have 3-4 KW of power per rack, which limits the number of highly compact servers that can be built into a rack. For the sensible use of high-density computing, however, a permanent power output of at least 10 - 12 KW is required in order to accommodate three times the number of servers on the same footprint. The new processor generations also have significantly more cores, which means that the increased use of virtualization enables a further improvement in relational performance/footprint.
The use of GPUs such as Tesla V100 can provide further performance boost with a moderate increase in power consumption. Graphics processors are thus a core component of high-density computing
However, it must be borne in mind that high-density computing requires additional floor space and effort to provide higher performance racks and compensate for the required additional cooling capacity.
The advantages of a high-density data center include:
High-Density Systeme | ||
---|---|---|
MicroCloud | MicroBlade | SuperBlade |
Area of application:
|
Area of application:
|
Area of application:
|
Features:
|
Features:
|
Features:
|
2U Twin | FatTwin | BigTwin |
Area of application:
|
Area of application:
|
Area of application:
|
Features:
|
Features:
|
Features:
|
GPU Server |
Area of application
|
||
Features:
|
All of the high-density systems we offer consist of several configurable server chassis that share a common chassis with a common power supply. The individual server slots for MicroCloud and Twin articles are called nodes and are the same for each article. However, each node can be configured individually when using the configurator (processor, memory, memory, software). If a configurator does not exist or is not used, additional information must be provided for a different configuration of the individual nodes. If the configuration is identical, there are no additional remarks. Blade systems consist of different, freely selectable enclosures and blades assigned to the enclosures. With the MicroBlade and SuperBlade family, two different blade families are available. The individual server inserts for blade servers are called blades. The blades are divided into groups and assigned to the blade types and different enclosures. In the configurator, the configuration can begin with the selection of a suitable enclosure or blade. If the configuration starts with an enclosure, the blades compatible with the selected enclosure can be selected and individually configured according to the desired type. If the maximum number of blades that can be installed has been reached, a warning is issued when you make a further selection. The selection of further blades is still possible, since spare blades are often bought as spare parts.
Do you want to configure your system yourself? Select the category that meets your needs and you will be directed to the configuration.
If you need additional information about High Density systems, you can find a detailed description of each system by following the links below: