It is no surprise that the demand for computing power is in dizzying increase, and the need for flexible and scalable software-defined data centers is equally significant.
In a large number of cases, applications are driven by data and artificial intelligence. This implies creating important workloads, as is normal for both artificial intelligence models, often accompanied by huge amounts of data needed for these apps to actually do their job.
The amount and diversity of data to be managed increases daily, and it is easy to imagine that the exponential increase in smart working, ecommerce and e-learning are playing a decisive role, as well as the increasing diffusion of IIoT and edge com
According to Chris Wright, Senior Vice President and Chief Technology Officer of Red Hat, it is necessary to run technology in synergy and work to meet customers’ needs. It becomes imperative to break the patterns linked to a monolithic view of data centers, evolving towards modern software models defined.
Data center evolution also passes through the adoption of new hardware components, and in this light the most promising technology is represented by data processing units (DPUs), which can have an impact on almost all levels of the IT landscape, whether they are individual systems
The DPUs combine multiple accelerators, making it possible to offload critical activities from the CPU to the dedicated hardware.
The DPU concept is not entirely new, but only recently the horizons have expanded decisively. Until recently, implementations focused on a particular function, such as network acceleration using SmartNIC.
The most structured DPU implementations combine a multi-core CPU that can be easily programmed with a evolved network interface and a powerful set of network accelerators, storage and security.
These advanced DPUs can be programmed to perform multiple software-defined functions and with hardware acceleration, exemplified by the recently launched NVIDIA BlueField-2 DPU.
By offloading on BlueField-2, IT organizations can achieve multiple goals, such as freeing the Cpu of a host server to run business applications and providing a dedicated platform for performing critical security and management functions.
This, in turn, leads to a software-defined data center that can be made up of optimal resource use and, as an additional advantage, provides additional visibility on how workloads are actually performed.
BlueField-2 offers developers encryption, network and hardware storage acceleration capabilities. Thanks to edge computing and high-performance storage systems, performance can be optimized while improving the use of the server.
According to Chris Wright, Red Hat and Nvidia see a bright future in which new software-defined data centers can be exploited and reconfigured with containers.
Red Hat plans to support BlueField-2 DPU with Red Hat Enterprise Linux and Red Hat OpenShift by collaborating with Nvidia to create data centers for the future.
Many Red Hat customers ask to move the IT workloads closer to the digital transformation edge on a large scale, require a reliable technological partner able to offer powerful, scalable and fully open software technologies to integrate hardware innovations.
Red Hat, says Wright, does not expect its customers to choose between the hardware needed and the fully open hybrid cloud portfolio, so the company emphasizes the provision of business-class open source solutions on multiple footprints and multiple arcs