Software
To install Microsoft Machine Learning Server offline, you’ll need the correct installation files—directly from Microsoft’s official sources—before you even think about deployment.
Skipping this step means wasted hours waiting for downloads to complete mid-setup, especially in remote environments. Below, I’ll walk you through the exact files required for Windows and Linux, where to get them legally, and how to verify them before you begin.
Microsoft Machine Learning Server installation files: required downloads for offline Windows/Linux deployments
Deploying Microsoft Machine Learning Server (ML Server) in restricted environments? You’ll need the right installation files before you begin. Without them, your offline deployment risks failing mid-setup due to missing dependencies or corrupted archives.
My guide covers the exact files you need for Windows Server and Linux (RHEL/CentOS/Ubuntu), including direct download sources and verification steps to ensure compatibility.
First, confirm your ML Server version and OS specs. For example, ML Server 9.4 requires Windows Server 2016/2019 or Linux RHEL 7.4+. Check Microsoft’s system requirements to avoid compatibility issues later.
I’ll walk you through downloading the core files—ML Server binaries, prerequisites, and dependencies—from official sources, then verify their integrity before transferring them offline.
Here’s the summary table of required installation files by OS, including direct download links and file sizes:
| Component | File Name | Download Link | Size | OS Support |
|---|---|---|---|---|
| ML Server Core | MLServer-Windows-x64.msi | Microsoft Download Center | 1.2 GB | Windows Server 2016/2019 |
| ML Server Core | mlserver-linux-x64.tar.gz | Microsoft Download Center | 950 MB | RHEL 7.4+/CentOS 7.4+/Ubuntu 18.04+ |
| Prerequisites | dotnet-runtime-2.1.5-win-x64.exe | .NET Core 2.1 | 120 MB | Windows |
| Prerequisites | dotnet-runtime-2.1.5-linux-x64.tar.gz | .NET Core 2.1 | 85 MB | Linux |
| Dependencies | R-3.5.1.tar.gz | CRAN | 2.1 GB | Linux (RHEL/CentOS) |
| Dependencies | Python-3.6.8.tar.xz | Python Official | 180 MB | Linux/Windows |
Next, verify the file integrity using checksums. Microsoft provides SHA-256 hashes for each download. For example, the MLServer-Windows-x64.msi file should match the hash listed in the release notes. Use PowerShell or Linux commands like sha256sum to confirm:
Step-by-step guide to transferring and preparing ML Server installation files offline
Transferring Microsoft Machine Learning Server installation files to an air-gapped system requires careful planning. Without internet access, you’ll need to download files on a connected machine first, then securely transfer them via USB drives or network shares.
Each step must account for file verification and dependency checks to avoid installation failures later. I’ll walk you through the process for both Windows Server and Linux (RHEL/CentOS/Ubuntu) environments.
Start by gathering all required files from Microsoft’s official download center. For Windows, you’ll need the ML Server installer, .NET Framework 4.7.2, and SQL Server dependencies.
On Linux, download the tar.gz archives for your specific OS version. Always verify file integrity using SHA-256 checksums provided by Microsoft to ensure no corruption occurred during transfer.
Step-by-Step Transfer and Preparation
- Download Files on Connected Machine: Use Microsoft’s official links to download the ML Server installation files, prerequisites, and checksum files. Store them in a dedicated folder (e.g., C:\MLServerInstall or /mnt/MLServerInstall).
- Verify File Integrity: Compare downloaded files against Microsoft’s SHA-256 checksums using CertUtil (Windows) or sha256sum (Linux). Example:
sha256sum MLServer9.4.1.tar.gz. - Transfer Files Securely: Copy files to a USB 3.0 drive or encrypted network share. For Linux, use dd to create a bootable USB if needed:
dd if=MLServerInstall.iso of=/dev/sdX bs=4M status=progress. - Extract Archives on Air-Gapped System: On Windows, right-click and select Extract All. On Linux, use
tar -xzvf MLServer9.4.1.tar.gz. Ensure extraction paths match your installation directory (e.g., /opt/microsoft/mlserver). - Configure Offline Repository (Linux): Create a local repo for dependencies. Example for RHEL/CentOS:
create-repo -p /var/www/html/mlserver-repo -v MLServer9.4.1 createrepo /var/www/html/mlserver-repo - Test Silent Installation (Dry Run): On Windows, use
/quietflag:setup.exe /quiet /norestart. On Linux, test with--dry-run:sudo ./mlserver-install --dry-run --prefix=/opt/microsoft/mlserver - Document All Steps: Record file versions, checksums, and installation paths for future reference. Store logs in /var/log/mlserver_install.log (Linux) or C:\Logs\MLServer (Windows).
For Windows Server, silent installations are straightforward with command-line flags. On Linux, you’ll need to configure yum/dnf repositories or use local RPM/DEB packages. Always test the offline repository on a non-production system first to catch dependency issues early.
Pro tip: Use PowerShell for Windows automation or Ansible for Linux deployments to reduce manual errors.
Remember, air-gapped systems are vulnerable to supply chain attacks. Always verify file sources and use digital signatures where possible. If you encounter permission errors during extraction, run the process as Administrator (Windows) or with sudo (Linux). Document any deviations from Microsoft’s official instructions for audit purposes.
By following these steps, you’ll minimize downtime and ensure a smooth offline ML Server deployment. Whether you’re setting up a remote lab or a secure production environment, this method guarantees you have all files ready before installation begins. Happy deploying! 🖥️