3 min

Tags in this article

, ,

OT security is a tenuous issue for the entire IT infrastructure. That doesn’t have to be the case, Nvidia believes.

We know Nvidia primarily as a GPU maker. As a driving force for the AI hype, the company has equipped countless data centres worldwide with the latest hardware. However, it has much broader ambitions than just these chips. Its software package focuses on more than just AI applications; it also provides extensive support for industrial settings, for example. Siemens has been featured several times at Nvidia keynotes, presenting realistic digital twins of cargo ships, factory halls and more. Those digital creations are made possible by combining Nvidia software and specialized hardware.

OT devices capture data that are extremely useful for industrial applications. For example, sensors can detect when preventative maintenance is needed before the device gets broken. Also, if equipped with a multitude of cameras and AI analytics, you can see the routes warehouse workers are taking and then have it analysed by AI to see how to improve efficiency. In short, there are plenty of applications that make it attractive to combine plenty of OT infrastructure with proficient hardware.

A“game-changing” approach

However, OT faces several challenges. An Internet connection is frequently out of the question to keep things like “smart” sensors and Industrial Control Systems (ICS) secure. If an internet connection is actually needed, organizations often choose to keep OT infrastructure air-gapped from the regular IT environment.

Tip: OT security of data centers should be a top priority

While that sounds more secure, that’s not necessarily the case. For example, it is virtually impossible to gain visibility into potential attackers looking to take over sensors and deploy them as a botnet. Because OT sensors primarily just run on Linux, hacks can follow a tried-and-true formula with conventional malware. If no one is watching for an OT compromise, hackers get a very easy ride indeed. Security company Sygnia, which collaborates with Nvidia, therefore finds that keeping OT separate greatly slows the detection of and response to cyber threats. So time for a “game-changing” approach.

Hardware-accelerated solutions at the edge of the network offer a way out. Edge processing involves separating the wheat from the chaff regarding data. A vast amount of measurements is already being gathered and analyzed so that OT monitoring, which would otherwise be separated from its IT equivalent, can remain visible. Compressing this data prevents IT environments from being disproportionately burdened as well.

The specific hardware solution is the Nvidia BlueField DPU (Data Processing Unit). This data card processes up to 400 Gigabits per second of data. Regarding software, Nvidia offers the Morpheus AI framework, which accelerates threat detection. BlueField cards are already a common find within conventional data centres, but at edge locations, they can efficiently process data and provide insights without additional .

Integration, integration, integration

As with generative AI applications, Nvidia is integrating on the security front. Partnerships with Fortinet, F5 Networks, Lacework and Palo Alto, among others, are already in place. Now, Sygnia is joining this ecosystem, linking Sygnia’s Velocity MXDR platform to Nvidia on the software front. So-called Pathfinder sensors from Sygnia additionally work seamlessly with Nvidia DPUs.

IT and OT come together in this way with specialized solutions. Where OT security is mostly considered a blind spot, this is precisely how it should be securely connected to IT infrastructure. The dedicated hardware also enables AI acceleration, which previously required your edge to connect to the rest of your network in a potentially insecure way. With Sygnia and Nvidia, this connection is limited to what is strictly necessary, while cybercriminals can no longer have their way with virtually undetectable botnets.