Navigating the Evolving Landscape of Network Management
The Current State of IT Operations
Recent reports from analysts at Forrester have highlighted a critical gap in IT departments: while they manage billion-dollar portfolios, their internal operations often lack the necessary automation and visibility. As network architectures grow increasingly complex, there’s a pressing need for network management to evolve. This need isn’t a recent development; it’s the natural progression of technology.
The Role of Automation and Machine Learning
Modern network management tools are now equipped with automation capabilities designed to streamline operations. Big data is increasingly utilized in Security Information and Event Management (SIEM), while machine learning (ML) empowers network administrators to foresee potential issues before they escalate into significant problems.
As Phil Huang, Business Development and Field Application Manager at D-Link, points out, “Our pure cloud management platform, backed by AI assistance, enables real-time monitoring and proactive alerts on potential problems.” This capability is transforming how network management is approached.
Shifting Skill Requirements
Automation doesn’t just simplify network management; it also transforms the skillsets required in this field. Matt Stava, CEO and Chairman of Spinnaker Support, notes that the demand for Cisco-certified experts is diminishing.
The skill set for modern network administrators is evolving. Amit Katz, Vice-President of Ethernet Switch at Nvidia, emphasizes this shift in his March 2026 blog post by noting that expertise in application programming interfaces (APIs) is growing in importance. The ability to use languages like Ansible, Salt, and Python is now more valuable than traditional certifications.
The New Landscape of Network Management
As network management evolves, the tasks that administrators need to perform have also changed. Katz highlights a transition from legacy tools that engage in polling devices using Simple Network Management Protocol (SNMP) to advanced switch-based telemetry models that actively stream diagnostic data.
He underscores a critical point: managing AI clusters requires a distinctly different approach than traditional data center management. “AI is not merely an extension of big data; it introduces complexities of its own,” he states.
How AI is Reshaping Network Operations
AI’s integration into network management is changing the way organizations troubleshoot and optimize networks. According to Marc Herren, a director at Information Services Group (ISG), AI can analyze vast amounts of data, identifying patterns that help streamline configurations to boost performance.
Industry leaders like Cisco and Juniper Networks are developing intent-based networking products. These technologies utilize AI to understand administrative intent, simplifying network configuration and management in the face of increasing complexity.
Complexity in Modern Networks
At Microsoft Build 2025, Phil Gervasi of Kentik articulated the challenges faced due to the growing intricacies of networks, which now span multiple clouds, data centers, and hybrid IT infrastructures. The influx of telemetry, events, and logs has surpassed human capacity to analyze in real-time. Under this strain, network teams are pressured to enhance issue resolution times and maintain uptime—often without expanding their workforce.
Gervasi clarified, “AI offers not magic but a more efficient way to correlate data and predict network behavior.” Through the use of AI, operators shift from reactive to predictive management, which can significantly improve operational efficiency.
The Promise of Large Language Models (LLMs)
While machine learning is making strides in capacity planning and anomaly detection, Gervasi also pointed to the potential of large language models (LLMs) in network management. Unlike traditional models that depend on structured data, LLMs can operate on unstructured data sources, such as documentation and configuration files. However, there are challenges, including the probabilistic nature of LLMs, which can lead to inconsistent outputs.
Gervasi emphasized the necessity for quality training data and effective evaluation strategies to ensure reliable performance of LLMs in network management.
Implementing AI in Network Management
Starting with AI in network management involves collecting various data sources: network telemetry logs, helpdesk tickets, and configuration files. These datasets need to be meticulously cleaned and formatted to make them accessible for AI systems.
Gervasi suggests using retrieval augmented generation (RAG) as a practical method to exploit this data. In this approach, user questions transform into mathematical representations that search for relevant data in a vector database, allowing the LLM to generate contextualized answers.
Additionally, text-to-SQL capabilities enable network engineers to utilize natural language for querying, simplifying their interaction with databases and offering graphical data representations.
The Future with Agentic AI
As the dialogue around AI continues, agentic AI emerges as a significant advancement. This type of AI mimics a human-like brain, assisting network management by autonomously coordinating tasks such as network tracing, data collection, and knowledge base consultations, ultimately generating remediation plans—all under human oversight.
Looking ahead, analyst Gartner forecasts that AI will be integrated into managed network services (MNS) by 2028, enhancing operational efficiency and empowering more informed decision-making. As organizations adapt to fluctuating demands, AI’s role will be pivotal in ensuring networks remain robust and agile.
In this rapidly evolving technological landscape, the ability to integrate AI into workflows is becoming a crucial skill for IT professionals. With increasing demands on network management accompanied by dwindling resources, mastering AI tools promises to enhance both operational efficiency and overall network performance.