kalc 0.1.5

Kalc - the Kubernetes Calculator core

kalc, the Kubernetes calculator

Overview

kalc is application aware Kubernetes rebalancing tool built on pure AI planning.

Quick Start

Requirements

• Linux x86_64
• Python 3.7+
• 6+ GB RAM, decent CPU
• Up to 20GB of disk space in /tmp for generated models
• kubectl installed and connected to cluster

Installation

pip install kalc

If your Linux host has other versions of Python installed you will recieve an error regarding "Could not find a version that satisfies the requirement kalc (from versions: )No matching distribution found for kalc"... If you recieve this, type the following to install kalc

python3.7 -m pip install kalc

Basic usage

$ kalc-optimize

kalc-optimize will generate bash scripts containing kubectl commands to get to more optimal states. Have a look at those scripts and execute any one of them, then stop and re-run kalc-optimize.

Autopilot

kalc can optimize your cluster in background, gradually increasing reliability by rebalancing and reducing cost by freeing nodes with low utilization. You can run kalc-optimize as a cron job, wait for X minutes and then run the most recent generated script file.

Architecture

• kalc-optimize will download current cluster state by executing kubectl get all and will start generating bash scripts into current folder
• Each generated bash script contains a sequence of kubectl commands to get the cluster in a more optimal state: better balanced nodes for availability and OOM/eviction resilience and a more compact packing
• As kalc continues to compute, it will emit more optimal states and bigger bash scripts with kubectl commands

kalc aims to take into account current policies, anti-affility, SLO levels and best practices from successful production Kubernetes clusters.

Project Status

kalc is a developer preview and currently supports a subset of Kubernetes resources and behaviour model.

We invite you to follow @criticalhop on Twitter and to chat with the team at #kalc on freenode. If you have any questions or suggestions - feel free to open a github issue or contact [e-mail] directly.

For enterprise enquiries, use the form on our website: kalc.io or write us an email at [e-mail]