MinIO AIStor is licensed under the MinIO Software License. An active license is required for production deployments. This page documents the installation and management of MinIO AIStor object storage using MinIO’s Red Hat OpenShift-certified operator through the Red Hat OLM. If you prefer to install on OpenShift using regular Helm charts, use the regular install instructions. This procedure assumes that the user interacting with the OpenShift cluster has authorization to:

MinIO AIStor is licensed under the MinIO Software License. An active license is required for production deployments. This procedure downloads and installs the AIStor Server onto bare-metal hosts running Ubuntu Server Linux. Review the Hardware and System Requirements to ensure your target infrastructure meets our guidance. You can also open a SUBNET issue to request an architecture review or assistance during deployment. MinIO strongly recommends using only Long-Term Support (LTS) Ubuntu releases in the LTS standard security maintenance lifecycle with a Linux Kernel version 6.8 or later, such as Ubuntu 24.04 LTS. See Linux Kernel Requirements for recommended distributions and performance details.

This section provides documentation and guidance for installing MinIO AIStor. Each subsection represents a supported platform, where guidance reflects the specific installation and configuration methods associated with that platform. If you need to install a key management service to manage cryptographic keys while objects are at rest, see the MinIO KMS docs. If you need to install DirectPV to manage container storage volumes in a Kubernetes environment, see the MinIO DirectPV docs.

Configure load balancers and reverse proxies for MinIO AIStor on Linux. Includes example configurations for NGINX and HAProxy.

Docs » Module code » h2o.model.metrics.binomial Source code for h2o.model.metrics.binomial import h2o from h2o.model ...

Object tiering is the moving of the data of an object from one set of hardware to another set of hardware. This is often done as a cost-saving measure, keeping more expensive and performant hardware for the newest or most accessed objects while placing other objects to less expensive and less performant storage solutions. Retrieve a tiered object When MinIO AIStor moves an object to another tier, the object’s metadata remains on the primary tier while the object’s data moves to the secondary tier. When MinIO AIStor needs to serve a tiered object, it transparently retrieves the object from the secondary tier.

MinIO AIStor supports deploying resources onto Windows hosts for local development and evaluation. For the purposes of evaluation or local development, MinIO recommends using a machine released in the last 4 years with at least 8 GB of RAM. You can use any supported version of Windows. For testing or local development using a virtualized environment, including Windows Subsystem for Linux (WSL), use a machine with a higher number of cores and 16 or 32GB of RAM.

MinIO AIStor is licensed under the MinIO Software License. An active license is required for production deployments. This procedure downloads and installs the AIStor Server onto a host running Microsoft Windows Server or Windows 11 Professional Edition. Review the Hardware and System Requirements to ensure your target infrastructure meets our guidance. Procedure Download the AIStor Server binary Invoke-WebRequest https://dl.min.io/aistor/minio/release/windows-amd64/minio -OutFile C:\minio.exe Create a data directory for use with the server The following example uses the path $HOME/minio/data. Changes this path to reflect the folder, storage volume, or drive intended for use with the server.

MinIO AIStor supports deploying resources as a container for local development and evaluation. Setting up MinIO AIStor as a container involves these steps: Deploy MinIO AIStor as a container Set up network encryption (optional) Performance The following table shows areas that have the greatest impact on MinIO AIStor performance, listed in order of importance.

MinIO AIStor supports compressing objects to reduce disk usage. Objects are compressed on PUT before writing to disk, and uncompressed on GET before they are sent to the client. This makes the compression process transparent to client applications and services. Depending on the type of data, compression may also increase overall throughput. Write throughput for a production deployment is generally 500MB per second or greater per available CPU core in the system. Decompression is approximately 1 GB per second or greater for each CPU core.