kubify 9003.1.8

Automated EKS

Welcome!

Kubify OS, Version: 9003.1.8

Well hello there! Welcome to Kubify. The Turn-Key DevOps/MLOps OS Developer-First Stack.

Who is this for?

Data Scientists/Devs/DevOps who want self service.

To rapid develop many services/models fast simultaneously on full real environment.

Why?

Because Docker-Compose and Terraform are 2 different tools, so I fixed it.

First class rapid testing, all your services listening for folder changes, so you can code fast, really fast.

If it works on your laptop, it works in prod.

How?

DevEx First Class Ideology

Usage?

See USAGE.rst

Infra Diagram?

See terraform/README.rst

Magic? Yes. Pure Magic.

This is a python package and a docker image (multi-arch).

1. PyPi

2. DockerHub

Cloud:

1. make cloud cloud=aws

2. make cloud cloud=gcp

3. make cloud cloud=azure

You have a redundant cloud!

Usage of published PyPi “kubify” package in your repo:

1. pip install kubify

2. optional services and/or terraform folders (default it uses the ones in this repo)

3. kubify start-all or kubify start service [] for rapid code development (all services listening for code changes on real environment)

Enjoy Rapid Testing!

Contributing:

(choose any to contribute to the kubify python package in this repo) A. devcontainer or container B. tox C. make docker D. make pip E. make pythons F. make test G. and more ..

Happy Coding!

Who are you?

I have a ton of MLOps and DevOps experience. I want to build an open source turn-key MLOps/DevOps stack, developer-first, self-service and redunant, as well as lowest cost (arm/spot) and scalable.

I am looking for contributors to build Kubify OS into a full-force turn-key DevEx solution.

Please contribute!

We all want self service turn key. It has arrived. No more K8s pains. Please contribute.

The future of DevOps is here. If only there was a make devs happy, well there is, welcome!

TODO: Implement DR Automation

Everything you build in Kubify has redundancy (dr, backups and active-active) and scalability in 2 regions, at the lowest possible cost (multiarch, spot, atom processor, spot gpu rapid de-scaling, mlops smart scaling, redundant scalers and more).

This is what true turn key feels like. DevOps in 1 day. Developer friendly. Purpose built for Data Scientists and Machine Learning, as well as Developers building services and cloud. Autopilot for DevOps, so your DevOps team can focus on company goals.

Kubify has been approved by AWS.

To use the python package

You can import kubify (we ship to PyPi), and then you can override any folder (such as terraform or services folders) by having folder with the same name in your repo.

See ./examples/simple for Python Package usage example.

To run this repo directly

To contribute to Kubify Open Source (and I hope you do), then clone this repo and:

1. make cloud cloud=[aws|gcp|azure]

2. make local

3. make local start-all

4. make local [service]

All the dependant services will start and will listen for code changes as well. Efficiently edit the entire codebase.

To install directly on workstation

To install directly on your workstation (instead of just using the container):

1. apple: make mac

2. ubuntu, debian and other debian-based: make deb

3. rhel, centos and other epel-based: make epel

Environment Isolation

Security Minute: You should not give prod access to anyone (but devs need to debug urgent issue in prod). How?

cd services/example-node-complex-svc
make local prod

What Just Happened?

You are hot patching many services locally (and with the full real environment) without access to data, but with the full real prod! Kubernetes can be amazing, and it is, with Kubify..

Contributing Workflow Patterns

1. DevContainer

2. Invoke Python Directly (pull image if missing tools)

3. Tox (multi-python testing)

4. Install Directly and Invoke Python Directly