Nornir-Edda

This repository implements the architecture proposed on A Network Programmability Essay. Which aims the following problems:

Abstraction of multiple vendors syntax in a single data model;
Decoupling of input data and configuration logic;
Single entry point for the configuration deploy;
Automatic vendor logic selection;

Requirements

Python 3.6.2 or higher
git
virtualenv
pip Additional requirements going to be installed automatically with pip

Installing

git clone https://github.com/renatoalmeidaoliveira/Nornir-Edda.git
cd Nornir-Edda/
virtualenv .
source bin/activate
pip install -r requirements.txt

Repository Structure

.
├── controllers						# Controllers folder
│   ├── Controller.py				# Controller Abstract Class
│   ├── Fabric.py					# Controller Fabric
│   ├── ios							# ios vendor Controller folder
│    │   ├── iosController.py		# ios Controller
│    │   ├── templates				# ios Jinja2 template
│    │    │   ├── interfaces.j2		# ios Interfaces template
├── models							# Models folder
│   ├── Model.py					# Model abstract Class
│   ├── interface.py				# Interface concrete Class
│   ├── interfaces.py				# Interfaces concrete Class
├── data							# Hosts data files
│   ├──cisco						# Device with hostname cisco data folder
│    │   ├── interfaces.yaml		# interfaces data model
│   ├──mikro						# Device with hostname mikro data folder
│    │   ├── interfaces.yaml		# interfaces data model
├── defaults.yaml					# Nornir defaults file
├── hosts.yaml						# Nornir hosts file
├── groups.yaml						# Nornir groups file
├──main.py							# Sample python scrips
├──requirements.txt					# Python pip requirements
├──NOTICE							# License Notice
└── README.md						# Readme file

Getting started

In this version only interfaces model for Cisco IOS, HP ComWare and MikroTik Routeos were implemented. For testing this repository follow the steps bellow:

Edit hosts.yaml file to suit your scenario
Create one folder for each host in your hosts.yml, in the data folder, where the folder name is the hostname, e.q ./data/hostname/
For each host create an interfaces.yaml with the desired configuration
Run main.py

Demonstration

Extending the models

If you want to deploy different models, follow the steps bellow:

Create a concrete class of the Model abstract class that will receive the data and output in some standard form
Implement the getModel() method that MUST return a string with same value as the Jinja2 template name.
Implement the getModelData() method that will return the data used by the Jinja2 template
For each vendor create the appropriate Jinja2 template with the same name returned by getModel(), i.e, f”{getModel()}.j2}”
Edit main.py to read the data model