CCNP Automation Bundle: Complete Learning Guide 2

Preparing for the new 350-901 AUTOCOR v2.0 exam — the core of the CCNP and CCIE Automation tracks? This complete, up-to-date guide covers all four exam domains in one volume, rebuilt for the 2026 blueprint version including the brand-new AI in Automation domain. Written for engineers who know the network but are newer to automation tooling, every topic starts from first principles and builds to exam-level depth — with real, runnable code you can try in a lab or in Cisco Modeling Labs.

Inside you'll master:

• Network Automation — Ansible, Terraform, RESTCONF/YANG, and Python (Netmiko, NAPALM, ncclient)

• Infrastructure as Code — Git, GitLab CI/CD pipelines, CML, Docker Compose, source of truth, YANG→JSON/YAML

• Operations — model-driven telemetry, logging, diagnostics, pyATS validation, TLS certificates, secure coding

• AI in Automation — AI-assisted dev risks, LLM security, building an MCP server with FastMCP, conversational agents, evaluating AI accuracy

What's included:

• Beginner-friendly explanations with worked examples and real code

• Summary + Key Takeaways for every topic

• 220 Cisco-style practice questions with fully explained answers

• The complete official exam blueprint and a glossary of key terms

Note: This is an independent, unofficial study guide and is not affiliated with or endorsed by Cisco Systems, Inc.

Every topic. Cisco's order. Cisco's words. Nothing else.

This book follows the published DCNAUTO 300-635 v2.0 blueprint exactly — five sections mapping onto the five domains, twenty-six sub-sections mapping onto the twenty-six topics. When Cisco writes "Construct an Ansible playbook with controller and device collections", there is a sub-section called that, and it is about exactly that.

That is not a formatting choice. It means when you finish a sub-section you can tick a line on the blueprint and know precisely what you have covered — and, more usefully, precisely what you have not.

• All 26 blueprint topics, weighted as Cisco weights them: Foundation (15%), Infrastructure as Code (25%), Programmability (25%), Operations (25%), AI in Automation (10%)
• 26 original diagrams, drawn for this book — no borrowed artwork, no vendor logos
• 130 knowledge-check questions — five at the end of every sub-section
• A 60-question practice exam, timed to 90 minutes and weighted to the blueprint, with a per-topic diagnostic
• A 13-exercise lab guide — every lab runnable on the free Cisco DevNet Sandbox, Containerlab, or a plain Linux VM
• A 69-term glossary built around the distinctions the exam actually tests
• Runnable code throughout — real playbooks, templates, plans and scripts, not fragments

Every sub-section carries a Common Pitfall (the mistake you were otherwise going to make), an Exam Tip (what the exam is likely to ask, and how it phrases it), and a From the Field section (why the topic matters at 3am, when it stops being a syllabus item).
You will learn why state: overridden deletes the VLAN you did not list. Why unreachable and failed are different halves of the stack. Why sample-interval 30 is thirty milliseconds. Why feature grpc and feature telemetry are not the same feature, and why enabling the wrong one produces silence. Why two Aggregator IDs mean LACP never formed. Why bridge vlan show is the command nobody runs and the answer surprisingly often.

There are only five ideas in this book, and the last chapter says so plainly. Declare the destination, not the journey. Idempotency is the safety property. Compute the change, show it to a human, then apply it. Every escape hatch trades structure for reach. Name the layer before you touch anything.

The approval gate appears four times — ansible-playbook --check --diff, terraform plan, Nexus Dashboard's Preview, and an AI agent proposing rather than applying. A candidate who memorises those as three separate facts is beaten by one who noticed they were one fact, because the second candidate can answer a question about the fifth implementation, which has not been invented yet.