Documenting Catalyst Center and SD-WAN Architectures at Scale

Large network transformations are easiest to understand when they're told in three states: where the network started, how it moved, and where it landed. That's exactly how I documented this one.

The deployment spans nine sites across India, EMEA, and the Americas — Mumbai and Chennai as the India hubs with Bangalore, Delhi, and Noida branches, London and Frankfurt in EMEA, New Jersey and Dallas in the Americas. Here is the journey, state by state.

01. Where we started

A classic enterprise WAN: every site connected over MPLS in a hub-and-spoke pattern, each router configured by hand over CLI, and the campus network and WAN managed as two separate worlds.

02. The management foundation

The design puts two management planes in place before any site migrates, each with a clear job. Catalyst Center automates the campus side — SD-Access fabric provisioning, software image management, and assurance analytics — working with ISE for identity and policy. The SD-WAN control plane runs the overlay: SD-WAN Manager for configuration and monitoring, vSmart for policy and routing, vBond for zero-touch onboarding.

03. The migration

With both planes up, sites moved in waves — a pilot branch first, then region by region, hubs last with their fabric handoff. Both transports ran side by side during the journey, so every site kept a clean path back until its wave was proven.

04. Segmentation, end to end

The achievement I'm most pleased with: a policy decision made in the campus fabric stays intact across the WAN. Each SD-Access virtual network hands off at the hub border to its own SD-WAN VPN — the same mapping at every hub, documented once.

05. Where we landed

The post-deployment state: nine sites on an application-aware overlay across MPLS, dual Internet, and LTE backup, with hub campuses running SD-Access fabric — and the whole system visible from two dashboards instead of dozens of terminal windows.

Where AI fits in

A deployment like this produces over a hundred documentation sections across two guides. I bring the topology decisions, the two-plane management design, and the handoff details, and Claude helps me draft at scale — the per-region variations, the repeated site procedures, and the cross-references between chapters. Every page then passes a strict build check before anything gets published.

Explore the full documentation

Each state above unfolds into complete, navigable documentation — the discovery registers, architecture and control plane chapters, segmentation and handoff design, configuration templates, wave procedures, and the operations pack. Both guides are built as illustrative enterprise scenarios, so the structure and decision logic are fully visible for anyone who wants to learn from them or adapt them.