Juniper Networks added more intelligence to its artificial intelligence (AI) networking platform to better steer outcomes for overworked network managers.
The updates build on the initial AI-Native Networking Platform that Juniper launched in early 2024. That initial launch included Juniper’s Marvis Minis technology, which brought a digital twins approach to Juniper’s AIOps model.
Christian Gilby, senior director of AI-native networking product marketing at Juniper Networks, explained that the updates are geared toward increasing trust in the use of AI to help direct network management.
The first part of the update is an extension of its Marvis Minis platform from a client on a managed device into the cloud. This means the ability to link insights from the Marvis Minis client that runs on a device into a cloud environment and act as an application insight platform that can identify and resolve network issues before they impact an end user.
“Let's say you're running a cloud application and that cloud service becomes unavailable. We're actually testing and validating that before a user hits the problem,” Gilby said. “Before, we focused more on the network, so we were focusing on the wired, the wireless, the WAN network. Now it's extending that to the application.”
The second major update is to the Marvis Actions dashboard, which now taps into the Marvis AI Assistant to provide more control over how “self-driving network operations” are enabled, a detailed history of all proactive actions, and insights into how the AI Assistant identified and resolved issues. This plays into what Gilby said was increasing the platform’s ability to gain trust of a network manager on the road from being a “driver-assist” platform to being a full self-driving option.
“A new capability here is I'm able to go in and for the full self-driving ones, I can either authorize or de-authorize Marvis to do that so that if [the network manager is] not yet comfortable having the network run itself, they can leave it in driver-assist mode,” Gilby explained. “They can come into this dashboard, audit when it's found a problem, what was the problem, and basically understand did it root cause the problem or was it a false positive, so that ultimately they start to turn on more self-driving and trust the network to run itself.”
The final update is to the Marvis Client, with the ability now to embed that client onto an end-user device. This can provide more real-world details on actual network performance.
“Today we can see what the AP sees,” Gilby said. “This will give us the ability to now see what a client sees. Maybe there's asymmetric where the client can't hear the access point, or vice versa, and now be able to understand what OS is running on the client or driver firmware, because a lot of times our customers find that can be causing issues where we're now able to detect these patterns. Maybe it's seven devices are running this firmware version ‘x’ and that's introducing the problem. Now we'll have that telemetry to be able to do that.”
The client update will initially support devices running Android, laptop and mobile versions of Microsoft Windows, and MacOS on the desktop. Gilby added that Juniper is working through possible support for Apple’s iOS.
More trust in AI
The updates follow on past moves by Juniper and continue its big push for greater trust in AI to fulfill its self-driving network mantra. Gilby acknowledged that this remains a challenge today.
“Early days, there was a lot of hesitation,” Gilby said. “There was a lot of customers who were thinking it was going to be false positives, where the AI would think there's a problem and there wasn’t one. I think by just experience deploying it, seeing it running, and where it would raise the action, not necessarily go and fix it for you, that's given us a couple years of history with them.”
Gilby did add that the technology is not there yet in terms of support for full self-driving networks but noted that Juniper is trying to be “open with our efficacy, when it detects a problem versus not.”
“In terms of the self-driving, we have a few capabilities there. I think there's more to come,” Gilby said.
Juniper’s AI focus under the HPE cloud
That more to come backs Juniper’s ongoing AI-focused platform development, which is happening under the cloud of its pending acquisition by Hewlett Packard Enterprise (HPE). That deal is currently tied up in a legal fight with the Department of Justice (DOJ), which is trying to block the $14 billion acquisition based on anti-competitive concerns.
Analysts have tied some of that concern to Juniper’s overarching Mist network platform, which is seen as a hearty rival to HPE’s recently updated Aruba platform. Those two platforms power their respective owner’s current position near the top of the overall wireless LAN market.
Woo Jin Ho, senior industry analyst at Bloomberg Intelligence, previously told SDxCentral that Mist was potentially the “biggest bone of contention” in holding up the deal.
“HPE needs a data center switching business, a good one, and there aren't many out there,” Ho said. “If we go down the line of the data center switching and routing businesses, the options are few and far between.”
Ho pointed to Cisco, Nokia, and Juniper in the routing space, and for virtual routers, “which HPE is trying to get into, maybe they could acquire someone like DriveNets or someone along those lines. But that’s more complimentary than anything else.”
That uncertainty is driving differing outcomes. Some analysts have pointed to current HPE and Juniper customers holding back on future purchase plans until the deal gets resolved, while HPE and Juniper management have downplayed any operational impact.
“Interest is strong,” Gilby said. “I think the differentiation has been the big thing. I think people … customers see with that potential acquisition, HPE sees strength in the Juniper Mist solution. I think, if anything, I would expect it to get invested in even more as time goes on whether or not the acquisition happens. We're just focusing on delivering this, and that's what I told my team, it's driving the business.”
