Awesome
AIStore on Kubernetes
AIStore is a lightweight, scalable object storage solution designed for AI applications. This repository serves as a complete toolkit for setting up AIStore in a Kubernetes environment, accommodating both managed Kubernetes services and bare-metal Kubernetes setups.
Overview
This repository includes a variety of resources to facilitate your deployment:
- Documentation/Guide: This section provides detailed, step-by-step instructions for deploying AIStore on Kubernetes (K8s), covering essential deployment scenarios and considerations.
- Ansible Playbooks: These playbooks are designed to streamline the setup of Kubernetes worker nodes for hosting AIStore deployments.
- Kubernetes Operator: AIS K8s Operator simplifies critical tasks such as bootstrapping, deployment, scaling, graceful shutdowns, and upgrades. It extends Kubernetes' native API, automating the lifecycle management of AIStore clusters.
- Helm Charts: [In development]. Helm charts for deploying AIS resources to be controlled by the operator (alternative to ansible).
- Monitoring: This guide provides detailed instructions on how to monitor AIStore using both command-line tools and a Kubernetes-based monitoring stack.
A Simple System Overview
The diagram illustrates a AIStore deployment on Kubernetes spread across multiple nodes, each containing a proxy
and a target
pod. The proxy
routes client requests to the target
pods, which handle data storage and retrieval. These pods utilize Persistent Volume Claims (PVCs) linked to Persistent Volumes (PVs) corresponding to actual storage disks. The AIS Operator oversees the entire setup, managing all operations related to the cluster.
Small Scale Experimental Deployments
This repository mainly focuses on production deployments of AIStore with multiple nodes each with multiple drives. If you don't require such scale then consider checking out the different deployment options available.
Deployment Guide
To successfully implement a multi-node deployment of AIStore in a production environment, thorough planning and strategic configuration decisions are essential. We recommend reviewing our Key Deployment Scenarios to determine the specific needs and objectives for your cluster. For a clear and detailed roadmap, our Step-by-Step Deployment Guide provides extensive instructions and best practices for setting up AIStore clusters on Kubernetes.
AIStore Operator
The AIS Operator is a key component in the ais-k8s system. It helps manage everything in an AIStore cluster, making tasks like starting, deploying, adjusting size, shutting down smoothly, and updating easier. It effectively handles AIStore resources within Kubernetes, adding to the Kubernetes API to fully automate the AIStore's lifecycle.
Important: Our deployment guide focuses on using the AIStore Operator for an easy and integrated setup process.