Installing machine learning components without internet access

Because the R and Python components provided with SQL Server 2016 and SQL Server 2017 are open source, Microsoft does not install R or Python components by default. Instead, we provide the related installers and bundled packages as a convenience on the Microsoft Download Center and other trusted sites. You must consent to the appropriate license, and then SQL Server setup installs R or Python components for you.

This topic provides the download locations for the installers and an overview of the offline setup process.

Overview of the offline installation process

Typically, setup of the machine components used in SQL Server 2016 and SQL Server 2017 requires an internet connection. When SQL Server setup runs, if you have selected any of the machine learning options, setup checks for the Python or R installers, as well as any other required components.

  • If the computer has an internet connection

    SQL Server locates and downloads the components for you, and then installs them during setup. Accept the license terms separately for each open source component (R or Python) that you install.

  • If the computer does not have internet access

    You must download additional installers before continuing with setup. At minimum, download the R or Python installers that are supported for the version of SQL Server that you are installing.

    Depending on your server's configuration, you might need additional components. See Additional components for details.

    After you have downloaded the installers, you use them when installing the feature as part of SQL Server setup.

Step 1. Obtain additional installers

For R

The R language is supported in both SQL Server 2016 and SQL Server 2017. Two different installers are required, for open source and proprietary components. The SQL Server setup wizard ensures that they are installed in the correct order.

  • Installers with SRO in the name provide the open source components.
  • Installers with SRS in the name contain components provided by Microsoft, including those for database integration.

For Python

The Python language is supported only in SQL Server 2017. Again there are two separate installers that you must download.

  • Installers with SPO in the name are for Microsoft Python Open, and provide the open source components.
  • Installers with SPS in the name are for Microsoft Python Server, and contain components provided by Microsoft, including those for database integration.

How to download

  1. Download the installers from the Microsoft Download Center sites onto a computer with internet access, and save the installer rather than running it.
  2. Copy the installer (CAB) files to the computer where you intend to install machine learning components.
  3. In SQL Server 2016, the setup wizard installed English by default. To install using a different language required modification of the installer file name, as described here: Modifications required for different language locales. For SQL Server 2017, the correct language is identified based on the instance locale.
  4. Download any additional components that are required, such as MPI or .NET Core.
  5. Optionally, you can download the archived source code for the open source components, but this is not required for SQL Server setup, and can be completed at any time. For more information, see R Server for Windows.

For a step-by-step walkthrough of the offline installation process for R Services in SQL Server 2016, we recommend article by the SQL Server Customer Advisory Team. the article includes screenshots and also covers patching and slipstream setup scenarios.

Step 2. Run offline setup using the SQL Server setup wizard

  1. Run the SQL Server setup wizard.
  2. When the setup wizard displays the licensing page, click Accept.
  3. A dialog box opens that prompts you for the Install Path of the required packages.
  4. click Browse to locate the folder containing the installer files you copied earlier.
  5. If the correct files are found, you can click Next to indicate that the components are available.
  6. Complete the SQL Server setup wizard.
  7. Perform the required post-installation steps to make sure the service is enabled.

Where to download machine learning components

Note

Be sure to get the files that match the version of SQL Server you are installing.

Support for Python is provided beginning with SQL Server 2017 CTP 2.0. Earlier versions, including SQL Server 2016, do not support Python.

Downloads for SQL Server 2017

Release Download link
SQL Server 2017 CTP 1
Microsoft R Open SRO_3.3.0.16000_1033.cab
Microsoft R Server SRS_9.0.0.16000_1033.cab
SQL Server 2017 CTP 1.1
Microsoft R Open SRO_3.3.2.0_1033.cab
Microsoft R Server SRS_9.0.1.16000_1033.cab
SQL Server 2017 CTP 1.4
Microsoft R Open SRO_3.3.2.100_1033.cab
Microsoft R Server SRS_9.0.2.100_1033.cab
SQL Server 2017 CTP 2.0
Microsoft R Open SRO_3.3.3.0_1033.cab
Microsoft R Server SRS_9.1.0.0_1033.cab
Microsoft Python Open SPO_9.1.0.0_1033.cab
Microsoft Python Server SPS_9.1.0.0__1033.cab
SQL Server 2017 RC1
Microsoft R Open SRO_3.3.3.22_1033.cab
Microsoft R Server SRS_9.2.0.22_1033.cab
Microsoft Python Open SPO_9.2.0.22_1033.cab
Microsoft Python Server SPS_9.2.0.22_1033.cab
SQL Server 2017 RC 2
Microsoft R Open SRO_3.3.3.23_1033.cab
Microsoft R Server SRS_9.2.0.23_1033.cab
Microsoft Python Open SPO_9.2.0.23_1033.cab
Microsoft Python Server SPS_9.2.0.23_1033.cab
SQL Server 2017 RTM
Microsoft R Open SRO_3.3.3.24_1033.cab
Microsoft R Server SRS_9.2.0.24_1033.cab
Microsoft Python Open SPO_9.2.0.24_1033.cab
Microsoft Python Server SPS_9.2.0.24_1033.cab
SQL Server 2017 CU1
Microsoft R Open use previous
Microsoft R Server SRS_9.2.0.100_1033.cab
Microsoft Python Open use previous
Microsoft Python Server SPS_9.2.0.100_1033.cab
SQL Server 2017 CU2
Microsoft R Open use previous
Microsoft R Server use previous
Microsoft Python Open use previous
Microsoft Python Server use previous

Downloads for SQL Server 2016

Important

When installing SQL Server 2016 SP1 CU4 or SP1 CU5 offline, download SRO_3.2.2.16000_1033.cab. If you downloaded SRO_3.2.2.13000_1033.cab from FWLINK 831785 as indicated in the setup dialog box, rename the file as SRO_3.2.2.16000_1033.cab before installing the Cumulative Update.

Release Download link
SQL Server 2016 RTM
Microsoft R Open SRO_3.2.2.803_1033.cab
Microsoft R Server SRS_8.0.3.0_1033.cab
SQL Server 2016 CU 1
Microsoft R Open SRO_3.2.2.10000_1033.cab
Microsoft R Server SRS_8.0.3.10000_1033.cab
SQL Server 2016 CU 2
Microsoft R Open SRO_3.2.2.12000_1033.cab
Microsoft R Server SRS_8.0.3.12000_1033.cab
SQL Server 2016 CU 3
Microsoft R Open no change; use previous
Microsoft R Server no change; use previous
SQL Server 2016 CU 4
Microsoft R Open SRO_3.2.2.13000_1033.cab
Microsoft R Server SRS_8.0.3.13000_1033.cab
SQL Server 2016 CU 5
Microsoft R Open no change; use previous
Microsoft R Server no change; use previous
SQL Server 2016 CU 6
Microsoft R Open no change; use previous
Microsoft R Server SRS_8.0.3.14000_1033.cab
SQL Server 2016 CU 7
Microsoft R Open no change; use previous
Microsoft R Server no change; use previous
SQL Server 2016 SP 1
Microsoft R Open SRO_3.2.2.15000_1033.cab
Microsoft R Server SRS_8.0.3.15000_1033.cab
SQL Server 2016 SP 1 CU1
Microsoft R Open no change; use previous
Microsoft R Server no change; use previous
SQL Server 2016 SP 1 CU2
Microsoft R Open SRO_3.2.2.16000_1033.cab
Microsoft R Server SRS_8.0.3.16000_1033.cab
SQL Server 2016 SP 1 CU3
Microsoft R Open no change; use previous
Microsoft R Server no change; use previous
SQL Server 2016 SP 1 CU4 and GDR
Microsoft R Open no change; use previous
Microsoft R Server SRS_8.0.3.17000_1033.cab
SQL Server 2016 SP 1 CU5
Microsoft R Open no change; use previous
Microsoft R Server no change; use previous

If you would like to view the source code for Microsoft R, it is available for download as an archive in .tar format: Download R Server installers

Additional prerequisites

Depending on your environment, you might need to make local copies of installers for the following prerequisites.

Component Version
Microsoft AS OLE DB Provider for SQL Server 2016 13.0.1601.5
Microsoft .NET Core 1.0.1
Microsoft MPI 7.1.12437.25
Microsoft Visual C++ 2013 Redistributable 12.0.30501.0
Microsoft Visual C++ 2015 Redistributable 14.0.23026.0

Installing for different language locales

If you download the .CAB files as part of SQL Server setup on a computer with internet access, the setup wizard detects the local language and automatically changes the language of the installer.

However, depending on your version of SQL Server, you might need to take additional steps to install the localized R components on a computer without internet access.

  • For SQL Server 2016

    After you download the R installers to a local share, you might need to manually edit the name of the downloaded files to insert the correct language identifier for the language you are installing.

    For example, to install the Japanese version of SQL Server, you would change the name of the file from SRS_8.0.3.0_1033.cab to SRS_8.0.3.0_1041.cab.

    Important

    This issue applies only to early versions and was fixed in later releases. Only use this workaround if the installer returns a message that it cannot install the correct language.

  • For SQL Server 2017

    Download the .CAB file for the R or Python components.

    The language is detected based on server locale. The correct locale is automatically installed using the downloaded .CAB file.

Slipstream upgrades

Slipstream setup refers to the ability to apply a patch or update to a failed instance installation, to repair existing problems. The advantage of this method is that the SQL Server is updated at the same time that you perform setup, avoiding a separate restart later.

  • If the server does not have Internet access, you must download the SQL Server installer, and then download matching versions of the R component installers before beginning the update process. The R components are not included by default with SQL Server.

  • If you are adding these components to an existing installation, use the updated version of the SQL Server installer, and the corresponding updated version of the additional components. When you specify that the R feature is to be installed, the installer looks for the matching version of the installers for the machine learning components.

Command-line arguments for specifying component locations

When performing an offline setup from the command line, you must provide the following command-line arguments to specify the location of components that you downloaded in advance. However, you do not need to set any additional flags to install additional required components; prerequisites such as .NET core are installed silently by default.

Location of installers

  • /UPDATESOURCE to specify the location of the local file containing the SQL Server update installer
  • /MRCACHEDIRECTORY to specify the folder containing the R component CAB files
  • /MPYCACHEDIRECTORY to specify the folder containing the Python component CAB files

R components in SQL Server 2016

  • /ADVANCEDANALYTICS to get engine support for external scripts
  • /IACCEPTROPENLICENSETERMS="True" to accept the separate R licensing agreement

R components in SQL Server 2017

  • /ADVANCEDANALYTICS to get engine support for external scripts
  • /SQL_INST_MR to use R
  • /IACCEPTROPENLICENSETERMS="True" to accept the separate R licensing agreement

Python components in SQL Server 2017

  • /ADVANCEDANALYTICS to get engine support for external scripts
  • /SQL_INST_MPY to use Python
  • /IACCEPTPYTHONLICENSETERMS="True" to accept the separate Python licensing agreement

Note

You cannot change the service account for Launchpad by using parameters in SQL Server setup. We recommend that you install using the default service accounts, and then modify the service account using SQL Server Configuration Manager. After doing so, be sure to restart the Launchpad service.

See also

Install Microsoft R Server

This article by the R Services Support team demonstrates how to perform an unattended install or upgrade of R services in SQL Server 2016: Deploying R Services on Computers without Internet Access.