359 votesunder review · AdminSQL Database feature voting forum admins (Product Owner, Microsoft Azure) responded
This is currently under review (meaning we see the benefit, but don’t have a time frame to share just yet). ThanksErik A. Brandstadmoen supported this idea ·Erik A. Brandstadmoen commented
For us that is on the SQL Azure platform, the capabilities when compared to SQL on-prem are very limited when it comes to Machine Learning support. SQL Server 2016 got lots of R capabilites, where you can run R scripts in-database. Now, with SQL Server 2017, Python support is added too. But we don't see this functionality being introduced in SQL Azure.
We are building more and more machine learning into our solultion, and using R, exposes via a REST endpoint. Today, we use AzureML for this, which is OK, but not really the best solution when it comes to a pure R solution.
It would be great for us if we could run our R scripts in-database in SQL server, as I understand you can now do in SQL Server on-prem, to avoid the complexity when it comes to error handling, community support (the intersection of people using AzureML and people making enterprise application is extremely small), speed (AzureML prediction is dead-slow compared to e.g. local dev box with Microsoft R Open), etc.