HPE Aruba Networking  is continuing to grow its capabilities with a series of new artificial intelligence (AI)-powered network security controls and observability tools announced this week at the RSA Conference in San Francisco.

Among the updates are new AI capabilities for enhanced IoT security in HPE Aruba Networking Central. The new AI-powered IoT discovery and classification features provide visibility into all connected devices and continuously monitor for rogue behavior.

HPE Aruba also announced new firewall-as-a-service (FWaaS) capabilities within its secure access service edge (SASE) solution. FWaaS extends firewall protection to data and devices across any environment without additional appliances. Additionally HPE Aruba is now integrating secure web gateway (SWG) into its EdgeConnect SD-WAN portfolio. This turnkey solution delivers comprehensive web threat protection for all users and devices, including unmanaged IoT.

HPE Aruba Networking has also done a lot of work to bring its Axis Security technology, which it acquired in 2023, deeper into the portfolio.

“We have integrated EdgeConnect SD-WAN and Aruba SD-Branch directly into Axis, with the ability to auto-learn the nearest tunnel endpoints, detect and route around failures, and automatically configure policies,” Jon Green, HPE Aruba Networking chief security officer, told SDxCentral. “This is the starting point and there are more to come.”

Who needs firewall as a service?

The new FWaaS offering isn't for everyone.

Green said that for organizations that are already using HPE Aruba Networking campus infrastructure, SD-WAN, or Axis, they don't need this firewall capability because it is already built into those services.

He said that FWaaS is to let customers cover other use cases, such as a non-Aruba SD-WAN product or a legacy routed network.  The main use case is someone who adopts the Aruba secure service edge (Axis) solution and starts using that to control policy around enterprise users accessing private applications, but who still has legacy use cases that don't fit into SSE.  FWaaS gives them a place to route network traffic that is destined for the cloud but came from a site without native firewall capability.

“Honestly this service is something we would expect to decline over time as customers go native ZTNA [zero trust network access], but it helps them transition,” Green said.

It's not genAI that HPE Aruba Networking is using for IoT security

While there is no shortage of hype in generative AI technologies in 2024, that's not the only way machine learning and AI technologies can benefit security.

Green said that for the new IoT discovery and classification capabilities, HPE Aruba Networking is using machine learning not generative AI.  He noted that HPE Aruba Networking has telemetry baked into its network infrastructure, feeding it to HPE Aruba Networking Central.

Green added that HPE Aruba Networking is able to build machine learning models on top of all that telemetry data for multiple purposes. It can be used to identify clusters of devices on the network, to identify the devices and to identify anomalies.

“To make this work you need widespread telemetry capture, which is why baking it into the network infrastructure makes the most sense, and you need lots and lots of data across lots of different customers,” he said. “HPE Aruba Networking Central gives us both.”

How HPE Aruba Networking is working with AI/ML

According to Green, the most difficult and critical part of any AI/ML development is having data.

HPE is starting from a good position on that front, Green said, with massive amounts of data coming from customer networks and stored in HPE Aruba Networking Central.

“We already leverage this data to build machine learning models that do things such as recommending configuration changes, pinpointing troublesome client devices, troubleshooting problems, etc,” he said. “We refer to this as AIOps, and there is a significant team of engineers and data scientists who continually release new capabilities in this area.”

He noted that an example of how this works is an “insight” which says “Your Wi-Fi network could achieve 12% improved throughput if you enabled 802.11ac, but you have 9 client devices with outdated driver software which has known compatibility problems with 802.11ac.  Here is the list.”  The source data for those recommendations comes from examining configuration and performance data across over 100,000 customers and millions of devices.

Going forward, Green said that today’s hype is all around generative AI, which is best thought of in his opinion as a plain language interface into a huge dataset. HPE announced an AI search capability back in March that makes heavy use of GenAI.

“We’ll continue building out this capability as well so that network admins can have conversations with their network config and monitoring data rather than needing to create customized reports or know where a particular config knob is located,” Green said. “We think there’s a lot of future potential here so this is a big area of growth for us.”