| The name of the base system (platform): | Artificial intelligence (AI, Artificial intelligence, AI) |
| Developers: | Nvidia |
| Date of the premiere of the system: | 2021/03/09 |
| Technology: | Data Mining, ITSM -, Network Health Monitoring IT Service Management Systems - Network Monitoring or IT Infrastructure Health-Performance Management, Virtualization, Network Application Performance Management Systems |
Content |
The main articles are:
- IT Service Management (ITSM)
- Virtualization. Classification and applications
- Data mining Data mining
2024: Discovery of a dangerous vulnerability in the Nvidia Base Command Manager management system
At the end of November, FSTEC sent out a vulnerability warning BDU:2024-10174[1]found in the NVidia Base Command Manager (BCM) workload management and infrastructure monitoring product, which is part of NVidia AI Enterprise. This tool optimizes cluster deployment, manages workload, and monitors infrastructure built on NVidia products. The danger of vulnerability is estimated at 9.8 out of 10 in CVSSv3. The vulnerability has been confirmed by the manufacturer and has already been fixed.
The BDU:2024-10174 error is due to the lack of an authorization procedure in one of the NVidia BCM components - the CMDaemon daemon. As a result of exploiting this vulnerability, a remote attacker can execute arbitrary code. Versions from 10.24.08 to 10.24.09 are vulnerable. The bug is fixed in version 10.24.09a, to which the company recommends updating.
NVidia BCM provides administrators of large data centers where NVidia devices are installed with all the tools for deploying and managing the artificial intelligence data center. The presence of such an error gives attackers the opportunity to gain control over the huge computing resources that are now used to train ML models and neural networks.
 | Vulnerabilities in Base Command Manager pose a serious risk, "said TAdviser, Sergey Gordeychik CEO of". " CyberOK- Unfortunately, such vulnerabilities are really common and occur in the Internet environment. As for NVidia, CyberOK specialists previously identified vulnerabilities on ML servers. DGX The vendor's approach pleasantly surprised: the identified vulnerabilities were promptly eliminated, and clients whose servers were on the Internet were notified in a timely manner. In Russia, NVidia BCM is mainly used only in large installations due to the high cost. We do not know about the copies available from the Internet. |  |
If the implementation of the main recommendation - updating the software to the corrected version - cannot be implemented, then FSTEC experts strongly advise to take the following compensatory measures:
- Use firewall tools to restrict remote access to the vulnerable software;
- Apply "white" lists of IP addresses to restrict access to the vulnerable software;
- Restrict remote access to vulnerable software using virtual private networks.
To these recommendations, we can add that the data center security service, where such a management tool is installed, should for some time more closely monitor the behavior of the corresponding software component - CMDaemon, trying to identify signs of exploitation of the vulnerability in its logs. This can be done both using compromise indicators (IoC) and analyzing the behavioral activity of the application.
2021
Integration with VMware vSphere 7 Update 2
On March 19, 2021, VMware introduced an updated version of VMware vSphere 7 Update 2 with AI-Ready Enterprise support. Read more here.
Nvidia AI Enterprise Announcement
On March 9, 2021, the company NVIDIA announced NVIDIA AI Enterprise, a software suite of enterprise tools and frameworks artificial intelligence optimized, certified and supported by NVIDIA in 7 VMware vSphere Update 2.
According to the company, as part of an industry collaboration to develop the AI-Ready Enterprise platform, NVIDIA teamed up with VMware for virtualizations AI-related work tasks VMware in vSphere using NVIDIA AI Enterprise. This enables enterprises to develop a wide range of AI solutions, such as advanced diagnostics in, health care smart enterprises for manufacturing, and service fraud detection financial.
With the NVIDIA AI Enterprise software suite, IT professionals from hundreds of thousands of enterprises using vSphere to virtualize computing can now support AI with the same tools they use to manage large-scale data centers and hybrid cloud environments. The NVIDIA software package provides scalable, multi-node performance of AI applications in vSphere, indistinguishable from bare servers.
| | Until now, users have run AI on dedicated physical servers (bare-metal servers). NVIDIA AI Enterprise allows customers to reduce the development time of AI models from 80 to 8 weeks, and allows them to deploy and manage advanced AI applications in VMware vSphere with the same scalable NVIDIA accelerated computing performance that is possible on a dedicated server. narrated by Justin Boitano, vice president and director of corporate and edge computing at NVIDIA | |
| | Each company is exploring how to upgrade its infrastructure to meet the requirements of artificial intelligence applications. With NVIDIA AI Enterprise and vSphere 7 Update 2, VMware users can now quickly track AI in their virtualized data centers and deploy a certified AI-enabled infrastructure for their modern applications. | |
One of the first to adopt NVIDIA AI Enterprise for vSphere was Optum Technology, part of the UnitedHealth Group.
| | AI plays an important role in the data-driven services Optum provides to United Healthcare. With NVIDIA AI Enterprise software and infrastructure running in our VMware vSphere environment, we can support the workloads of our applications and deploy AI across the enterprise. supplemented by Justin Potuznik, Senior Chief Engineer, Optum Technology | |
NVIDIA AI Enterprise provides compatibility with a wide range of accelerated CUDA applications, AI frameworks, pre-models trained , and software development toolkits running in a hybrid cloud. Optimization allows you to scale workloads to multiple nodes and run large deep learning models with a full virtualization GPU.
To support applications in NVIDIA AI Enterprise, VMware vSphere 7 Update 2 is now certified for graphics processors kernels with NVIDIA A100 tensor in NVIDIA certified systems that include large from servers ,,, and Dell Technologies. HPE Lenovo Supermicro With this certification, vSphere customers purchasing a license for NVIDIA AI Enterprise receive direct user support from NVIDIA.
NVIDIA has also certified vSphere for software computing virtualization, providing support hypervisor for dynamic migration with NVIDIA Multi-Instance GPU technology, which allows each A100 GPU to be divided into a maximum of seven instances at the hardware level to optimize the efficiency of workloads of all sizes.
In addition, some NVIDIA ConnectX adapters are now certified for VMware vSAN via RDMA (Remote Direct Memory Access), which removes CPU communication tasks and optimizes application performance and infrastructure ROI.
NVIDIA AI Enterprise software is available as a perpetual license for 3595 dollars USA per CPU week. Enterprise Business Standard support for NVIDIA AI Enterprise costs $899 per year for a license. Customers can apply for early access to NVIDIA AI Enterprise when planning an upgrade to VMware vSphere 7 Update 2.
Notes
| The site content is translated by machine translation software powered by PROMT. The machine-translated articles are not always perfect and may contain errors in vocabulary, syntax or grammar. Read original article If you find inaccuracies or errors in the results of machine translation, please write to editor@tadviser.ru. We will make every effort to correct them as soon as possible. |