Nvidia, Dell and VMware put AI and Zero Trust in their data center offer


Two of the world’s largest technology companies in their respective industries – VMware for enterprise software and Nvidia for high-end computing hardware – are joining their collective talents in the idea of ​​bringing conventional data centers into the era from the metaverse. To do this, they plan to integrate artificial intelligence and the “Zero Trust” security approach into the servers from the outset.

The two companies today announced at the VMware Explore 2022 conference in San Francisco a new data center platform they say designed for the age of AI that will standardize a disparate list of features – cutting-edge AI modeling, AI inference, data science and security capabilities at Ero Trust – in servers around the world.

The package combines Dell’s latest PowerEdge servers with Nvidia’s BlueField-2 DPUs (data processing units), GPUs and enterprise AI software, all optimized for the new Dell workload platform. Enterprise vSphere 8 from VMware, which was also showcased at today’s event. Enterprises can curate and distribute the functions of these elements on Nvidia LaunchPad, a convenient lab platform that provides access and control of hardware and software to build end-to-end AI workflows for data scientists , the companies said.

Safety is task number one

Improving security for these next-generation workloads using multiple firewalls is the primary strength behind this partnership project.

“With this package, we’re protecting workloads with next-generation firewalls,” Brad Doctor, vice president of security engineering at VMware, told reporters. “Now you will see firewalls in every server. So we are no longer confined to how previous data centers were built. »

“With this (conventional) model, you have a router or firewall at the edge of the data center. All traffic inside the data center was assumed to be reliable. Here we move to a model where we will place a next-gen firewall with NSX (from VMware) in each server rather than running it on a CPU, which will offload it to the GPU. So with that we can do things like intrusion detection and intrusion protection systems. »

The security system can be as granular as immediately recognizing that a potential intruder is typing a password too slowly when trying to gain unauthorized access to a server, Doctor said.

Data isolation is key

“Isolation is super important in all of this – it’s the best security in the world,” said Manuvir Das, head of enterprise IT at Nvidia. “Security has always been a focal point of the VMware NSX platform, and now hosting it on the Bluefield gives us a new layer of isolation between the application processing domain and the computing domain. infrastructure. »

According to Brad Doctor, running vSphere 8 on Bluefield DPUs dramatically speeds up production workloads. By offloading data to the DPU, users can speed up networking and security services while saving CPU cycles, he added.

Distributed applications with AI/ML and analytics are advancing the transformation of data center architecture by using accelerators and providing better security as part of common application infrastructure, Krish said. Prasad, senior vice president and general manager of VMware Cloud Platform Business, in a note to the press.

“Dell PowerEdge servers built on the latest VMware vSphere 8 innovations and accelerated by BlueField-2 DPUs deliver next-generation performance and efficiency for enterprise cloud applications while better protecting businesses against side-stream threats in multi-cloud environments,” said Krish Prasad.

Nvidia AI Enterprise, a suite of cloud-native AI and data analytics software, is optimized for businesses to use AI on familiar infrastructure. It’s certified to be deployed anywhere — from the enterprise data center to the public cloud — and includes global enterprise support to keep AI projects on track, Manuvir Das said.

Dell servers with vSphere 8 on BlueField-2 DPUs will be available later this year, Doctor said. Nvidia AI Enterprise with vSphere is now available and can be deployed using LaunchPad hands-on labs.

Source: ZDNet.com





Source link -97