Choosing the edition

SQL Server is no longer just a database, but has grown into a database platform - an ecosystem -  which consists of many additional services (such as SSRS, SSAS, and SSIS) that supports and also extends the capabilities of modern database usage. When installing Machine Learning R Services (in a database), one should think about the ecosystem environment and which additional services would be used along with R Services. If the business need requires advanced R (or Python) integration and analytics, the Enterprise edition is the right one. If only basic R integration is needed, the standard version will cover the needs. Also, think along the lines of other analytical tools if you need analysis services or reporting services, and which developments tools would also be needed for that (for example, MDX on top of OLAP cubes and running R code against the same data mart).

When you have decided on the version, download the ISO or CAB installation file of SQL Server 2017(or 2016) and start the installation. I will install the Developer edition of SQL Server 2017 (which is, from the installation perspective, almost the same as the 2016 version):

Installing a new installation of SQL Server will guarantee that the Machine Learning Services with R (or in-database service) are correctly installed.

In the next step, the installation features must be installed. To have R Services installed, a tick must be placed on R for SQL Server Services (in-database), while for SQL Server 2017, a tick must be placed on R in the Machine Learning Services (in-database) section:

In the following server configuration step, you must check the users and accounts that will have access to services. For R Services (in-database), the SQL Server Launchpad service will be installed and automatically started. This service can be started or stopped after the installation through Windows Application-Services:

After the database engine is configured, you will be prompted to accept the agreement to install Microsoft R Open, making sure that you are aware of the fact that R is under GNU License. By requesting this consent, Microsoft just wants to make sure that the administrator agrees and that all new updates and patches to R Open version will be in accordance with the SQL Server update preferences:

In addition to agreeing to R's agreement, please note that the life cycle of Microsoft R Server is two years. If MRS 8.0 was released in January 2016, the official support ended in January 2018; to be more precise, security and critical updates will come in for a period of one year (until January 2017) and, after that, only the security updates will continue until January 2018. During the year, upgrades will also be received. Please note that it is relevant for the standalone product-Microsoft Machine Learning Server, but it is worth mentioning how long the support timeline will be held.

If you are also installing Python, the same consent will be requested:

Once you have selected all of the features, configurations, and consents, you will be presented with an overview of the services and features to be installed:

Once the installation is completed, you will have the R Engine for Machine Learning Services and Microsoft Machine Learning Server with R (if selected) installed. Please note that R Engine for R Services (in-database) will have a different R installation, as the standalone Microsoft R Server, and also all the installed packages will be different, under different paths, rights, and security settings.