Mist Systems’ AI-Driven Wi-Fi Solution Takes Aim At Cisco, HPE Aruba

Hot on the heels of inking new partnerships with VMware and Juniper Networks, networking start-up Mist Systems is throwing the AI gauntlet at the feet of Cisco Systems and HPE Aruba by rolling out several enhancements to its AI-driven Wi-Fi solution.

"Aruba is making all their money selling controllers, they're not pushing the cloud at all," said Jeffrey Aaron, Mist vice president of marketing, told CRN. "Cisco has Meraki, which doesn't scale. They don't have a cloud solution that scales for the large enterprise and they don't have AI built into the core."

"Cisco is adding DNA [Digital Network Architecture] as an overlay box on top of their existing wireless controller boxes, and that's crazy," Aaron said. "Anyone who's doing AI, you're not doing it on a hardware appliance, you're doing it in the cloud where you have infinite compute and infinite storage. It's not a scalable, resilient or cost-effective solution to have to put in all these boxes at $75,000 a pop on top of your controllers. They understand the need for it, they just don't have the architecture, and that opens it up for us."

One executive at HPE Aruba, Vice President and General Manager of Wireless and Software Ash Kumar Chowdappa, refuted Aaron's arguments in a statement to CRN: "Aruba offers a comprehensive cloud management and service assurance solution driven by artificial intelligence and machine learning for optimizing network performance and availability and ensuring a great end-user experience. Aruba enables customers to embrace the network architecture that is best for their business needs and choose either cloud or on-prem manageability options for managing the wireless LAN, wired and SD branch infrastructure portfolio it offers."

"That doesn't stand up to reality," said a Cisco executive in response to Aaron's claims. "It's an extremely bold claim to say Meraki doesn't scale. We have 12 years of experience at Meraki running cloud-based architecture and running it at scale. We have a global hotel chain running 23,000 access points. We have a major health provider running Meraki in all its kidney dialysis centers, so that's a real life or death scenario. More than 25,000 Meraki devices can be managed effectively through a single login."

"DNA isn't only a wireless controller, it's a central controller for the entire network," the Cisco executive continued. "We have a long history of doing machine learning at Cisco. DNA Center does ML in the box. Our customers want hybrid architectures. DNA Center is a software. The only reason we sell the appliance is to control the environment people run their controller software in. Today, it only supports [Cisco] UCS, which is why it looks like an appliance model. In the future, it'll be available on other hardware platforms and be available as a VM."

Mist's AI-driven wireless LAN is an industry first and stands alone in the market, Aaron said. The Cupertino, Calif., start-up developed its own access points that feed into the cloud without any extra hardware or sensors in the overlay, he said. "AI is built into the core to do things like machine learning, even correlation for trouble shooting, natural language processing, anomaly detection. It makes Wi-Fi more predictable, more measurable for things like location services, asset visibility and analytics."

Now, the company is making several AI enhancements intended to make networks seamless and self-healing, Aaron said.

Mist is making enhancements to its Marvis virtual network assistant, which uses natural language processing to help troubleshoot Wi-Fi networks. The AI-based system has been learning since it was launched early this year, and Aaron said it is now ready to answer hundreds of specific questions concerning individual network devices from phones to access points.

The API-driven Mist Learning WLAN submits help desk tickets and notifications automatically when Marvis detects anomalies, allowing IT folks to solve problems before they even know they exist, the company says.

Mist's Wi-Fi Assurance Service collects information about Wi-Fi coverage, capacity, throughput and performance. Now, the service can be customized to measure the impact of changes being made as a result of the information being collected. The company says the capability marks the first time that wireless networks can detect problems, learn from them, automatically make changes using AI and then determine how effective those changes were.

Mist's Wi-Fi Assurance service comes with WAN SLEs to cover capacity, coverage, throughput, latency, access point uptime and roaming. The company is now allowing IT administrators set, monitor and enforce those SLEs.

The enhancements to Mist's platform come as the company deepens relationships with some of the networking market's key players. Mist inked a partnership agreement with Palo Alto Networks in May, with VMware's VeloCloud unit last month, and with Juniper Networks last week.

Together, the group has formed the AI for IT Consortium, which Aaron describes as "a bunch of like-minded vendors coming together to put best-of-breed solutions together and bring automation across the whole stack."

As a result, customers can get "a seamless wired and wireless experience without having to be locked into a single vendor like Cisco or HPE," Aaron said, although he didn't rule out working with Cisco's Viptela SD-WAN unit in the future. Both Viptela and Mist count Verizon as a customer.

The opportunity for solution providers, especially MSPs, is huge, Aaron said.

"For our MSP partners or resellers, this is just more insight, more automation to lower Wi-Fi opex and ultimately deliver better Wi-Fi experience to end-users," Aaron said. "We're encouraging partners to use our APIs to do their own innovation, as well. We have partners that are doing it: Ecosystem integration, integration with help desk systems, value-added services. We're all about having an open, programmable interface that lets our resellers and MSPs come in and add value to the equation by customizing the services they offer."

"Our best partners right now are the ones that are thinking ahead of the curve," Aaron said. "They're looking to provide more value. This is an opportunity for them to differentiate on the wireless side and bring in some of those other portfolio companies."

Jonathan Strong, co-founder of Technology Engineering Group, a Medina, Ohio, MSP that works with Mist, as well as Cisco Meraki, said Mist is "changing the wireless networking landscape."

"The client-up approach to wireless and AI allows our customers for the first time to take a proactive posture for their wireless networks," Strong said. "No more on-premise controllers or cloud-hosted controllers that make it nearly impossible to see the details of what is going on inside the network. Now we have insights, analysis and remediation before our clients even detect any issues."