These days, networks aren't just racks of switches anymore, they're also AI clusters. A single training job depends on a lossless fabric, a Kubernetes control plane, a cloud VPC, an inference gateway, and a telemetry pipeline that has to notice trouble before anybody files a ticket. Even with something as strong as Python glue, it's not going to hold up at that size. But C++23 doesn't. When you get to the end of this book, you'll have a fully-fledged automation tool at your disposal.
The book kicks off with std::expected, std::span, ranges, and coroutines applied to real device work, then builds a CMake and vcpkg project with sanitizers, tests, and a containerlab fabric. Then you can drive NETCONF, YANG, and gNMI, provision servers over SSH and Ansible, program OVSDB and VXLAN overlays, automate Suricata and Vectorscan detection, and reach AWS and Azure through one interface. The last couple of chapters show you how to use Kubernetes resources on the server side, attach pods to RDMA devices through Multus and SR-IOV, and validate PFC and ECN across a RoCEv2 fabric. Plus, there'll be token streaming for vLLM replicas, your tooling will be exposed as an MCP server, and the loop will be closed with OpenTelemetry and guarded remediation. Everything's on one Linux laptop.
Key Learnings
Plan every change so it can be undone, and see it before it runs.
Hand back errors as values with std::expected, so nothing fails quietly overnight.
Read packets without copying them, using std::span, and let sanitizers catch your mistakes.
Check five hundred devices at once with coroutines, and never write a callback.
Check a change against the device model before you send it anywhere.
Roll out to a few devices first, and widen only when they behave.
Let your C++ drive Ansible, Suricata, and ovs-ofctl instead of rebuilding them.
Apply Kubernetes changes safely, then check from inside the pod that they worked.
Send AI requests to the least busy replica, and tune the connection for tokens.
Spot trouble with limits, moving averages, and a model that advises but never acts.
Table of Content
C++23 for Network Automation
Setting up Automation Environment
Writing, Testing, and Validating Automation Scripts
Device Configuration with NETCONF, YANG and gNMI
Server Provisioning and Configuration Management
Software-Defined Networking
Practical IDS/IPS Automation
Cloud Network Automation
Kubernetes as AI Cluster Control Plane
RDMA and GPU Fabric Automation
Target Readers
If you're a network engineer, systems programmer, or part of an infrastructure team, then this book is for you. You don't need to have any prior experience with SDN, Kubernetes, or machine learning for this book. Just some basics of coding around C++ and fundamentals of networking is sufficient.
