Closed-Loop Network Automation: Architectural Principles from Catalyst Center-Scale Systems

Abstract

Enterprise network controllers must manage large, heterogeneous environments of devices, clients, locations, applications, and services while preserving explainability, reliability, and change safety. This paper proposes a vendor-neutral architecture informed by public descriptions of Cisco DNA Center and established work on intent-based networking, telemetry, assurance, and autonomic networking. The architecture organizes closed-loop operation into six interacting layers: data acquisition and context, correlation and diagnosis, intent and policy, workflow orchestration, outcome verification, and external integration. It also defines hierarchical control scopes, evidence aggregation, action classes, and control budgets to limit concurrent remediation and reduce operational risk. Cisco DNA Center provides product context; this paper does not claim access to proprietary implementation details. The paper discusses proposed assessment dimensions, including outcome coverage, telemetry freshness, diagnostic accuracy, control-loop convergence, failover resilience, and operator understanding. The framework offers a product-aware, vendor-neutral basis for analyzing scalable enterprise controller automation and assurance.