Getty Images/iStockphoto
Juniper adds data center networks to Mist AI
Juniper's Mist AI update will include using the Marvis virtual network assistant to obtain information on data center cabling, configurations and connectivity issues.
Juniper Networks has added the data center to its AI for IT operations platform for the branch and campus.
Juniper introduced the latest capabilities of its cloud-based Mist AI on Monday. The company also introduced two QFX switches and a PTX router for data center servers that run AI model inference and machine learning (ML).
The releases come less than three weeks after HPE reported it planned to acquire Juniper for $14 billion by early next year.
Mist AI uses big data analytics, ML and other AI technologies to identify and resolve common networking issues. Juniper takes Mist AI to the data center via software it acquired when it bought Apstra for an undisclosed sum in 2021. Juniper kept the Apstra name and continued developing its hardware-agnostic software.
"Extending Mist to Apstra adds a proven network-focused AIOps platform to a multi-vendor data center network automation tool," said Shamus McGillicuddy, an analyst at Enterprise Management Associates. "This should streamline, automate and optimize management of data center networks."
Apstra monitors hardware and software configurations. It also provides tools to determine the impact of configuration changes before deploying them. Companies providing similar capabilities include Arrcus, Forward Networks and Kentik.
Juniper rivals Cisco and Arista also offer AI tools to find and fix network issues across multiple networks from a single dashboard. However, the capabilities differ considerably based on each vendor's networking hardware and software.
Marvis VNA update
Juniper plans to update its Marvis natural language virtual network assistant in the second quarter. The release will let engineers obtain information on data center cabling, configurations and connectivity issues. Today, Marvis, a component of the Mist AIOps platform, can provide such information only on wired and wireless campus and branch networks.
Marvis will also suggest remedies to problems and use generative AI to provide information from data center product documentation and knowledge bases. Companies that decide to make Marvis-suggested network changes must make them through Apstra separately.
"Giving administrators a tool to extend visibility past the Juniper stack is simply not something we see out of a hardware vendor, as they're usually [focused on] their own equipment," said Scott Marrone, IT director for the city of Parkland, Fla., a Mist AI customer.
Scott MarroneIT director, City of Parkland, Fla.
Juniper plans to launch by the end of March a wireless network monitoring feature named Marvis Minis. The feature continuously performs simulation testing of wireless connections to all user devices to spot issues before an event disrupts service. Before the introduction of Minis, an event had to occur before the Mist AI platform reported it.
"This is [potentially] game-changing because it means the [IT] department, for the first time ever, can be proactive rather than reactive," Marrone said.
Juniper's latest QFX switches deliver 800 Gb Ethernet through the Broadcom Tomahawk 5 silicon. The new PTX router provides the same throughput with the Juniper Express 5 silicon.
Juniper has updated Apstra to provide congestion management, load balancing and flow control for AI and machine learning traffic on the new hardware. Also available for AI applications are validated designs accessible via Marvis.
The latest hardware supports remote direct memory access (RDMA) over Converged Ethernet version 2, an internet layer protocol for routing packets. RDMA encapsulates InfiniBand transport packets over Ethernet for server-to-server data movement between application memory.
The latest capabilities are in the QFX5230, available now, and the 5240, shipping by the end of March. Juniper plans to release the PTX10002 in the second half of the year.
Antone Gonsalves is an editor at large for TechTarget Editorial, reporting on industry trends critical to enterprise tech buyers. He has worked in tech journalism for 25 years and is based in San Francisco.