Getting Started
This section walks you from first download to a running Alluxio cluster with verified data access. Each installation guide is self-contained and covers cluster setup, storage mounting, and initial verification.
What You'll Deploy
An Alluxio cluster consists of three types of components:
Worker
Caches data on local NVMe/SSD; serves reads directly to clients
1+
Coordinator
Background jobs: cache loading, mount table management, worker health monitoring
1
etcd
Cluster membership and mount table storage
3 (production), 1 (evaluation)
Your applications connect to the cluster after installation via FUSE (local filesystem mount), S3 API (HTTP, S3-compatible), or Python FSSpec. Each access method is set up separately once the cluster is running — the installation guides cover the initial cluster only.
What You'll Need Before Starting
For hardware, OS, networking port, and resource sizing requirements, see Prerequisites.
Download link from Alluxio — includes Docker image
.tarfiles and a license string.Private Docker registry (Kubernetes only) — images must be pushed to a registry accessible from your cluster nodes before deploying. Not required for Docker / bare-metal, where images are loaded with
docker loadon each host.Storage backend ready — an S3 bucket (with credentials), GCS, HDFS, or other supported UFS. See Underlying Storage for the full list.
Choose Your Installation Guide
When to use
You have an existing Kubernetes cluster
Linux hosts, EC2 instances, or Slurm — no Kubernetes
Both paths produce the same cluster. The choice depends entirely on your infrastructure.
Cloud-specific guides build on top of the Kubernetes path:
OCI OKE — Oracle Cloud Infrastructure Kubernetes Engine, including VCN setup, OCIR image push, and
oci-bvworker page store.
AI-Assisted Deployment
AI-Assisted Deployment Guide — a Claude Code skill that asks about your environment and produces a personalized, step-by-step deployment plan. Useful if you're new to Alluxio or want guided help through the process.
Last updated