CloudVision and Automation
CloudVision (CV) is the management and telemetry platform for EOS fleets. Understand its layers: CloudVision Portal (CVP) for configuration management, change control, and compliance; CloudVision as a telemetry and analytics collector; and the CloudVision APIs for programmatic access. Know the change-control workflow: a change is built as a Change Control, which snapshots the intended config, runs a dry-run/diff, requires approval, and executes with rollback capability. This is the concept most often tested - CV is not just a dashboard; it enforces a reviewable, auditable change pipeline.
Know the automation toolchain: eAPI for direct device interaction, Ansible/CVP modules for orchestration, and CloudVision's configlet/tag-based provisioning. Tags and containers group devices so that configlets apply to matching devices automatically. Understand ZTP (Zero Touch Provisioning) and how a device bootstraps its configuration from a server.
Traps: (1) Assuming a configlet applies immediately without a change control - in managed mode, changes flow through change control. (2) Confusing compliance (drift detection against intended config) with reconciliation (actually pushing the fix). (3) Forgetting that CV telemetry is streaming, not polled SNMP. (4) Mixing up tags vs. containers as the targeting mechanism.
Rule of thumb: "review, approve, execute, roll back" = change control; "detect drift" = compliance. Worked example: to roll out an NTP change to all campus switches, tag the devices, attach a configlet to the tag, build a change control, review the diff, approve, and execute - with a one-click rollback if the diff is wrong.