Juniper Networks Announces AI-Native Networking Platform

Moving Wi-Fi troubleshooting to the cloud offers substantial benefits even before full AI implementation. So, data for troubleshooting can be readily available in the cloud. Techniques like dynamic packet capture improve the process and are more secure than traditional methods like secure socket shell (SSH). Once data is in the cloud, AI can diagnose complex issues by combining network and application-specific data.

The AI component in AI-native networking platforms is crucial for enhancing automation and improving experiences for organizations. AI represents an evolution from traditional algorithms to deep learning models, similar to technological advancements like ChatGPT. These AI models can predict end-user experiences with apps such as Zoom and Microsoft Teams, identifying specific features that contribute to better experiences.

A key advantage of AI in networking is its ability to pinpoint which part of the network path is causing customer issues, especially in complex environments where apps are distributed across private data centers, multiple public clouds, and edge environments. This ability to isolate problems in a higher context and then drill down to specific issues ensures that as apps move across various platforms, the user experience remains consistently high.

When Mist Systems transitioned to implementing AI in networking, it required significant architectural and organizational changes. The architectural component involved building a real-time cloud infrastructure from the ground up. The organizational shift was equally important, though often overlooked, which involved hiring a support team with networking expertise and placing it alongside the data science team.

A major consideration for organizations evaluating AI solutions is to determine whether a vendor employs AI for IT operations (AIOps) tools within its own support team. Incorporating AI into the vendor’s operational processes, beginning with the support structure, is a crucial initial stage in implementing cloud AIOps in networking. This method aligns with the “eating your dog food” philosophy, emphasizing the significance of using and trusting one’s own product.

Juniper Networks and Mist Systems, pre-acquisition, have effectively demonstrated that AI systems evolve and improve over time. Organizations must understand that AI systems may not be perfect initially, but their performance will be enhanced significantly as they learn and adapt. This ongoing improvement is a strong indicator of genuine AI use, distinguishing it from basic automation tools labeled as AI.

Networking is shifting away from traditional command-line interfaces (CLIs) and dashboards to natural language interfaces with intuitive, conversational network systems facilitated by generative AI. It not only enhances the user interface through conversational capabilities but also adds valuable tools to the data science toolbox for more effective network management.

Marvis Minis is a new tool in the Juniper AI-Native Networking Platform that acts like an automatic problem-solver. Integrated with the Mist AI engine, the tool takes network management from reactive to proactive, using machine learning (ML) to continuously analyze network activity and correct problems before they affect users. Marvis Minis eliminates the need to manage additional overlay networks. It is integrated directly into network elements like access points (APs), switches, and routers.

Juniper plans to extend Marvis cloud AIOps into broader areas, including application experiences and more complex network layers. The goal is to collect data from various sources, including the entire enterprise and service provider portfolios, as well as app-level data. The guiding philosophy here is that more data leads to more accurate and detailed answers, which drives the effectiveness of cloud AIOps.

The role of network engineering is undergoing significant changes. For many network engineers, AI is becoming a vital tool for practical use cases. The relationship between network engineers and AI systems is evolving into a collaborative one, where AI is a trusted and integral part of the team, like a virtual team member who helps solve problems.