- ServiceNow said network incident volume fell 90% after it moved its network operations to Juniper Mist AI
- Network patching dropped from 2,000 to 3,000 hours per year to fewer than 60, while the network team stayed the same size
- ServiceNow is aiming for a 100% autonomous network by 2028, built up through gradual handoffs of control
ServiceNow cut its network incident volume 90% after moving its network operations to HPE's Mist AI platform, ServiceNow's Senior Director of Information Systems Management Sajeev Nair said at a recent event.
Network patching that once took 2,000 to 3,000 hours a year now takes fewer than 60, and the company is aiming for a fully autonomous network by 2028.
Nair described ServiceNow's results on stage with Rami Rahim, EVP, president and GM of HPE Networking, at HPE's Networking Investor Day in Sunnyvale, California, on Sept. 30. ServiceNow is one of HPE's longest-running references for self-driving networks using AI. HPE said the system now resolves 70–80% of potential network issues on its own.
Overhauling the network
In 2017, ServiceNow's network was "chaotic," Nair said. The company bought point products to solve one problem at a time, each adding another control plane and splintering network telemetry further. Engineers had to correlate data by hand to find the source of a problem.
"The hidden element into this whole thing was the cost that was so high that we did not realize that there was a better way to solve this problem," Nair said.
In 2018, Mist Systems, then a startup, pitched ServiceNow on two ideas: measuring the end-user experience, which Nair said the networking industry had overlooked, and running AI across the entire wireless network. ServiceNow started with a small deployment.
Juniper Networks acquired Mist in 2019, which extended Mist's AI to Juniper's switching and routing. ServiceNow rebuilt its network during the COVID-19 pandemic.
"We are just basically going to rip and replace every single vendor stack that we had in our networking environment," Nair said.
The result was Wi-Fi, wired and WAN networking managed on a single platform, with Marvis, Mist's AI engine, running across all three. A 2023 Juniper case study described a 13-month migration covering 30 sites, using Juniper access points, EX switches and Session Smart SD-WAN routers.
What changed for ServiceNow's network team?
Along with the 90% drop in incidents, problems that once took hours to troubleshoot are now found in minutes, Nair said. Network patching that used to run for hours over weekends is fully automated, which accounts for the drop from 2,000 to 3,000 hours a year to fewer than 60.
Notably, ServiceNow's headcount did not change. The same team that ran the network before ServiceNow deployed Mist runs it today, Nair said. ServiceNow "has grown immensely over the last several years, but their IT team has not," Rahim said in an interview with Fierce after the session.
ServiceNow has also tied Mist into its own platform. The two companies worked as design partners to integrate Mist with ServiceNow's IT service management software. Juniper and ServiceNow later extended that work into a managed-services offering that Deutsche Telekom uses for its enterprise networking services.
Early on in the transition, the network team was skeptical, Nair said. Anything that went wrong at ServiceNow was blamed on the network first, and engineers worried that handing control to AI would make things worse.
ServiceNow addressed that with what Nair called graduated autonomy: turning over control one step at a time and continuing to move forward rather than reversing course. The team that doubted the technology has since "fully leaned in," Nair said, and now spends its time on higher-level work.
That approach matches how HPE sells the self-driving network: customers enable autonomous actions incrementally as their confidence grows. HPE has highlighted similar results elsewhere. The U.K. Ministry of Justice reported about a 75% reduction in service desk tickets after deploying HPE's automated network actions, which launched in May.
What does ServiceNow want next?
ServiceNow has set "a bold goal" of a 100% autonomous network by 2028, Nair said.
He compared his job to running a utility. "When you flip the switch, the light has to come on. When I turn the tap, the water has to flow," Nair said. When something does go wrong, the network should tell users it has a problem, say it is fixing itself and point them to a part of the campus where their Zoom call will work, he said.
Network latency will matter more as AI inference spreads from data centers to laptops and phones, Nair said. That puts more pressure on networks to run without interruption.
Marvis, the engine behind ServiceNow's results, is becoming the common AI engine for both Mist and Aruba Central, which HPE says it will keep as separate platforms while sharing AI and hardware underneath. The self-driving network was one of several pitches HPE made at the investor day, along with its first AMD Helios order, a $1.2 billion deal with Vultr.
Read more about HPE's self-driving network on Fierce Network:
HPE's self-driving network fixes 70–80% of issues on its own, Rahim says
HPE won't merge Aruba Central and Mist, but the platforms will share AI, applications and hardware
HPE lands first AMD Helios order in $1.2B Vultr deal
HPE targets ‘low-hanging fruit’ in self-driving networks push
HPE sharpens its AI edge, hones networking chops
ServiceNow aims to simplify roaming issues for NTT Docomo and StarHub