Merge pull request #2707 from MicrosoftDocs/release-sql2017

Release sql2017 rc2 into release-pubsql-prep

Author: craigg-msft

docs/advanced-analytics/what-s-new-in-sql-server-machine-learning-services.md

@@ -2,7 +2,7 @@
 title: "What's New in Machine Learning Services | Microsoft Docs"
 ms.custom: 
   - "SQL2016_New_Updated"
-ms.date: "07/14/2017"
+ms.date: "07/31/2017"
 ms.prod: "sql-server-2016"
 ms.reviewer: ""
 ms.suite: ""
@@ -22,62 +22,104 @@ In SQL Server 2016, Microsoft introduced SQL Server R Services, a feature that s
 
 In SQL Server 2017, machine learning becomes even more powerful, with addition of support for the popular Python language. Along with the support for new languages comes a new name: **Machine Learning Services (In-Database)**.
 
-## What's new in SQL Server 2017 Release Candidate 1
+Catch the latest announcement here! [Python in SQL Server 2017: enhanced in-database machine learning](https://blogs.technet.microsoft.com/dataplatforminsider/2017/04/19/python-in-sql-server-2017-enhanced-in-database-machine-learning/)
 
-SQL Server 2017 builds on Microsoft's commitment to making it easy to build and deploy machine learning solutions, with the following new features:
+## What's new in SQL Server 2017 Release Candidate 2
 
-- Improved and expanded Python libraries
-- Updates to realtime scoring, plus a new native T-SQL scoring function
-- Updates to Microsoft R
-- Installation of pre-trained models
-- New package management capabilities for the DBA
+Microsoft Machine Learning Server in SQL Server now provides comprehensive support for building and deploying machine learning solutions. Here are the highlights of this release:
 
-### Improved Python integration in SQL Server
+> [!IMPORTANT]
+> 
+> Machine learning services, including use of R or Python, are currently not supported when running SQL Server on Linux, or in Azure SQL database. Look for changes in a later release.
+> 
+> However, native scoring using the PREDICT function is currently supported in the Linux edition. 
+ 
+### In-database Python integration
+
+You can run Python in stored procedures, or execute Python remotely using the SQL Server computer as the compute context. This integration opens up new avenues for the vast community of Python developers and data scientists to use the power of SQL Server, and to explore innovations from Microsoft such as **revoscalepy** and **microsoftml**.
+
+SQL Server developers gain access to the extensive Python libraries from the open source ecosystem, including popular frameworks such as scikit-learn, Tensorflow, Caffe and Theano/Keras. 
+
+But running Python in-database isn't just for machine learning; there are a myriad of other potential applications for integrating Python with SQL, leveraging the strengths of the respective languages to deliver more intelligent, powerful solutions.
+
++ **revoscalepy**
+
+    This release includes the final version of **revoscalepy**, which supplies Pythonic equivalents of the scalable, streaming algorithms in RevoScaleR. You can create Python models for linear and logistic regressions, decision trees, boosted trees, and random forests, all parallelizable, and capable of being run in remote compute contexts.
+
+    For more information, see [What is revoscalepy](python/what-is-revoscalepy.md).
+
++ Remote compute contexts for Python
+
+    This release supports use of multiple data sources and remote compute contexts. The data scientist or developer can execute Python code on a remote SQL Server, to explore data or build models without moving data. Use of remote compute contexts requires **revoscalepy**.
+
++ Python support in Microsoft Machine Learning Server (Standalone)
+
+    SQL Server 2017 includes the option to install a standalone version of the Python and R platforms. By using Machine Learning Server, you can distribute and scale R or Python code without using SQL Server.
 
-Support for Python in SQL Server was introduced in CTP 2.0.  When you select Python as the language to install with SQL Server, you get the **revoscalepy** module, which supports the same scalable, distributed algorithms and compute contexts that are provided in RevoScaleR.
+    For an example of Python use in Microsoft Machine Learning Server, see [Publish and consume Python code](python/publish-consume-python-code.md).
 
-In this release, the **revoscalepy** library is joined by **MicrosoftML for Python**, which provides fast, scalable machine learning algorithms and data transformations that run on the same compute contexts and data sources supported by revoscalepy.
+### New algorithms
 
-For more information, see these topics:
+The **MicrosoftML** package for both R and Python contains state-of-the-art machine learning algorithms and data transformation that can be scaled or run in remote compute contexts. Algorithms include customizable deep neural networks, fast decision trees and decision forests, linear regression, and logistic regression.
 
-+ [Set up Python in Machine Learning Services](../advanced-analytics/python/setup-python-machine-learning-services.md)
+The MicrosoftML package comes with both R and Python interfaces, and is based on Microsoft Machine Learning Server version 9.2.0.
 
-+ [revoscalepy function reference](https://docs.microsoft.com/r-server/python-reference/revoscalepy/revoscalepy-package)
+For more information, see [Introduction to MicrosoftML](using-the-microsoftml-package.md) and [microsoftml for Python](https://docs.microsoft.com/r-server/python-reference/microsoftml/microsoftml-package).
 
-+ [MicrosoftML for Python](https://docs.microsoft.com/r-server/python-reference/microsoftml/microsoftml-package)
+### Operationalization
+
+This release contains multiple options and features to help you deploy and distribute machine learning tasks:
+
++ Operationalization with T-SQL
+
+    The integration of Python with T-SQL means that you can call any Python code using `sp_execute_external_script`. This secure infrastructure enables enterprise-grade deployment of Python models and scripts that can be called from an application using a simple stored procedure. Additional performance is by streaming data from SQL to Python processes and MPI ring parallelization.
+
++ **mrsdeploy** for Python
+
+    The **mrsdeploy** package for Microsoft R Server now supports deployment of Python models and scripts as web services, and is available as an option in Machine Learning Server (Standalone). For an example of how it works, see [Publish and consume Python code](python/publish-consume-python-code.md).
+
++ Performance
+
+    Microsoft has pushed the boundaries of performance for scoring. With in-database scoring, we processed a million rows per second using R models. In this release, new features for **realtime scoring** and **native scoring** support better performance in single-row scoring and small batches as well. 
 
 ### Realtime scoring and native scoring
 
-_Realtime scoring_ is supported for scoring in SQL Server and in Machine Learning Server (Standalone). This feature relies on native C++ libraries to read a model stored in an optimized binary format. From the model, the function can generate predictions without having to call the R runtime. This makes batch scoring much faster.
+Realtime scoring relies on native C++ libraries to read a model stored in an optimized binary format, and then generate predictions without having to call the R runtime. This makes scoring operations much faster.
 
-Additionally, this release of SQL Server 2017 includes a native T-SQL function for fast scoring that can be run on any edition of SQL Server, even on Linux. The function is available by default, and requires no installation of R or extra configuration. This means you can train a model elsewhere, save it in SQL Server, and then perform scoring without ever calling R. This feature is referred to as _native scoring_.
+Additionally, this release of SQL Server 2017 includes a native T-SQL function for fast scoring that can be run on any edition of SQL Server, even on Linux. The function requires no installation of R or extra configuration. This means you can train a model elsewhere, save it in SQL Server, and then perform scoring without ever calling R. This feature is referred to as _native scoring_.
+
+  - Native scoring is available only in SQL Server 2017. It uses a T-SQL function that can run in any edition of SQL Server, including Linux.
+ - Realtime scoring is supported in SQL Server 2017, and in Microsoft Machine Learning Server. You can runa  stored procedure or perform realtime scoring from R code.
+ - Realtime scoring is also available for SQL Server 2016, if the instance is upgraded to the latest release of Microsoft R Server.
 
 For more information, see these articles:
 
-+ [Realtime scoring](real-time-scoring.md)
-+ [Native scoring](sql-native-scoring.md)
-+ [How to perform realtime scoring or native scoring](r/how-to-do-realtime-scoring.md)
+ + [Realtime scoring](real-time-scoring.md)
+ + [Native scoring](sql-native-scoring.md)
+ + [How to perform realtime scoring or native scoring](r/how-to-do-realtime-scoring.md)
 
-## Other updates
+### Upgrade your ML experience and get pre-trained models
 
-This section lists features that were released previously, but that have been updated in this release.
+If you installed an earlier version of SQL Server 2016 R Services, you can now upgrade to the latest version by switching your server to use the Modern Lifecycle policy. By doing so, you can take advantage of a faster release cycle for R and automatically upgrade all R components. For more information, see [Microsoft R Server 9.0.1](https://docs.microsoft.com/r-server/whats-new-in-r-server).
 
-### Improved package management for data scientists
+The installer also offers the option to install a collection of pre-trained models in binary format. These models support machine learning in scenarios such as image recognition, where it might be difficult for customers to find large datasets to train a model. After you install one of the pre-trained models, you can use it for prediction on your own data without the time and expense involved in training such a large and complex model.
 
-To provide better support for databases administrators who must manage R package libraries, SQL Server now supports the CREATE EXTERNAL LIBRARY command.
+For more information, see [Install pre-trained models in SQL Server](r/install-pretrained-models-sql-server.md)
 
-For more information, see [R package management for SQL Server R Services](../advanced-analytics/r/r-package-management-for-sql-server-r-services.md).
+### Package management
 
-### Upgrade your R experience with pre-trained models
+This release includes many improvements in package management for SQL Server. These include:
 
-The RevoScaleR package is included in SQL Server 2016, SQL Server 2017, and Microsoft R Server. It includes transforms and algorithms that support distributed or parallel processing, and multiple compute contexts.
+- database roles to help the DBA manage and audit permissions
+- the CREATE EXTERAL LIBRARY statement in T-SQL
+- a rich set of R functions to help install, remove, or list packages owned by users.
 
-If you installed an earlier version of RevoScaleR with SQL Server 2016, you can now upgrade to the latest version by switching your server to use the Modern Lifecycle policy. By doing so, you can take advantage of a faster release cycle for R and automatically upgrade all R components. For more information, see [Microsoft R Server 9.0.1](https://docs.microsoft.com/r-server/whats-new-in-r-server).
+For more information, see [Package management](r/r-package-management-for-sql-server-r-services.md).
 
-Additionally, when you install the new R components, you can get a collection of pre-trained models in binary format. These models support machine learning in scenarios such as image recognition, where it might be difficult for customers to find large datasets to train a model. After you install one of the pre-trained models, you can use it for prediction on your own data without the time and expense involved in training such a large and complex model.
+### Get started
 
-For more information, see [Install pre-trained models in SQL Server](r/install-pretrained-models-sql-server.md)
++ [Set up Python in SQL Server Machine Learning Services](../advanced-analytics/python/setup-python-machine-learning-services.md)
 
-> [!IMPORTANT]
-> 
-> Machine learning services, including use of R or Python, are currently not supported when running SQL Server on Linux. Look for changes in a later release. However, native scoring using the PREDICT function is currently supported in the Linux edition.
++ [Set up R in SQL Server Machine Learning Services](r/set-up-sql-server-r-services-in-database.md)
+
++ [Machine learning tutorials](tutorials/machine-learning-services-tutorials.md)

docs/analysis-services/what-s-new-in-sql-server-analysis-services-2017.md

@@ -1,7 +1,7 @@
 ---
 title: "What's new in SQL Server 2017 Analysis Services | Microsoft Docs"
 ms.custom: ""
-ms.date: "07/17/2017"
+ms.date: "07/27/2017"
 ms.prod: "sql-server-2017"
 ms.reviewer: ""
 ms.suite: ""
@@ -18,6 +18,10 @@ manager: "erikre"
 # What's new in SQL Server 2017 Analysis Services
 [!INCLUDE[tsql-appliesto-ssvNxt-xxxx-xxxx-xxx](../includes/tsql-appliesto-ssvnxt-xxxx-xxxx-xxx.md)]
 
+
+## SQL Server 2017 Analysis Services RC2
+There are no new features in this release. Improvements in this release include bug fixes and performance.
+
 ## SQL Server 2017 Analysis Services RC1
 There are no new features in this release, however, this release includes additional improvements to [Dynamic Management Views](https://docs.microsoft.com/sql/analysis-services/instances/use-dynamic-management-views-dmvs-to-monitor-analysis-services) (DMV) for tabular models at the 1200 and 1400 compatibility levels.
 

docs/linux/quickstart-install-connect-docker.md

@@ -12,7 +12,7 @@ ms.assetid: 82737f18-f5d6-4dce-a255-688889fdde69
 ---
 # Run the SQL Server 2017 container image with Docker
 
-In this quick start tutorial, you use Docker to pull and run the SQL Server 2017 RC1 container image, [mssql-server-linux](https://hub.docker.com/r/microsoft/mssql-server-linux/). Then connect with **sqlcmd** to create your first database and run queries.
+In this quick start tutorial, you use Docker to pull and run the SQL Server 2017 RC2 container image, [mssql-server-linux](https://hub.docker.com/r/microsoft/mssql-server-linux/). Then connect with **sqlcmd** to create your first database and run queries.
 
 This image consists of SQL Server running on Linux based on Ubuntu 16.04. It can be used with the Docker Engine 1.8+ on Linux or on Docker for Mac/Windows.
 

docs/linux/quickstart-install-connect-red-hat.md

@@ -13,7 +13,7 @@ ms.assetid: 92503f59-96dc-4f6a-b1b0-d135c43e935e
 ---
 # Install SQL Server and create a database on Red Hat
 
-In this quick start tutorial, you first install SQL Server 2017 RC1 on Red Hat Enterprise Linux (RHEL) 7.3. Then connect with **sqlcmd** to create your first database and run queries.
+In this quick start tutorial, you first install SQL Server 2017 RC2 on Red Hat Enterprise Linux (RHEL) 7.3. Then connect with **sqlcmd** to create your first database and run queries.
 
 > [!TIP]
 > This tutorial requires user input and an internet connection. If you are interested in the [unattended](sql-server-linux-setup.md#unattended) or [offline](sql-server-linux-setup.md#offline) installation procedures, see [Installation guidance for SQL Server on Linux](sql-server-linux-setup.md).
@@ -52,11 +52,11 @@ To configure SQL Server on RHEL, run the following commands in a terminal to ins
    > Make sure to specify a strong password for the SA account (Minimum length 8 characters, including uppercase and lowercase letters, base 10 digits and/or non-alphanumeric symbols).
 
    > [!TIP]
-   > When installing RC1, no purchased licenses are required to try any of the editions. Because it is a release candidate, the following message appears regardless of the edition you select:
+   > When installing RC2, no purchased licenses are required to try any of the editions. Because it is a release candidate, the following message appears regardless of the edition you select:
    >
    > `This is an evaluation version.  There are [175] days left in the evaluation period.`
    >
-   > This message does not reflect the edition you selected. It relates to the preview period for RC1.
+   > This message does not reflect the edition you selected. It relates to the preview period for RC2.
 
 1. Once the configuration is done, verify that the service is running:
 

docs/linux/quickstart-install-connect-suse.md

@@ -13,7 +13,7 @@ ms.assetid: 31ddfb80-f75c-4f51-8540-de6213cb68b8
 ---
 # Install SQL Server and create a database on SUSE Linux Enterprise Server
 
-In this quick start tutorial, you first install SQL Server 2017 RC1 on SUSE Linux Enterprise Server (SLES) v12 SP2. Then connect with **sqlcmd** to create your first database and run queries.
+In this quick start tutorial, you first install SQL Server 2017 RC2 on SUSE Linux Enterprise Server (SLES) v12 SP2. Then connect with **sqlcmd** to create your first database and run queries.
 
 > [!TIP]
 > This tutorial requires user input and an internet connection. If you are interested in the [unattended](sql-server-linux-setup.md#unattended) or [offline](sql-server-linux-setup.md#offline) installation procedures, see [Installation guidance for SQL Server on Linux](sql-server-linux-setup.md).
@@ -53,11 +53,11 @@ To configure SQL Server on SLES, run the following commands in a terminal to ins
    > Make sure to specify a strong password for the SA account (Minimum length 8 characters, including uppercase and lowercase letters, base 10 digits and/or non-alphanumeric symbols).
 
    > [!TIP]
-   > When installing RC1, no purchased licenses are required to try any of the editions. Because it is a release candidate, the following message appears regardless of the edition you select:
+   > When installing RC2, no purchased licenses are required to try any of the editions. Because it is a release candidate, the following message appears regardless of the edition you select:
    >
    > `This is an evaluation version.  There are [175] days left in the evaluation period.`
    >
-   > This message does not reflect the edition you selected. It relates to the preview period for RC1.
+   > This message does not reflect the edition you selected. It relates to the preview period for RC2.
 
 1. Once the configuration is done, verify that the service is running:
 

docs/linux/quickstart-install-connect-ubuntu.md

@@ -12,7 +12,7 @@ ms.assetid: 31c8c92e-12fe-4728-9b95-4bc028250d85
 ---
 # Install SQL Server and create a database on Ubuntu
 
-In this quick start tutorial, you first install SQL Server 2017 RC1 on Ubuntu 16.04. Then connect with **sqlcmd** to create your first database and run queries.
+In this quick start tutorial, you first install SQL Server 2017 RC2 on Ubuntu 16.04. Then connect with **sqlcmd** to create your first database and run queries.
 
 > [!TIP]
 > This tutorial requires user input and an internet connection. If you are interested in the [unattended](sql-server-linux-setup.md#unattended) or [offline](sql-server-linux-setup.md#offline) installation procedures, see [Installation guidance for SQL Server on Linux](sql-server-linux-setup.md).
@@ -58,11 +58,11 @@ To configure SQL Server on Ubuntu, run the following commands in a terminal to i
    > Make sure to specify a strong password for the SA account (Minimum length 8 characters, including uppercase and lowercase letters, base 10 digits and/or non-alphanumeric symbols).
 
    > [!TIP]
-   > When installing RC1, no purchased licenses are required to try any of the editions. Because it is a release candidate, the following message appears regardless of the edition you select:
+   > When installing RC2, no purchased licenses are required to try any of the editions. Because it is a release candidate, the following message appears regardless of the edition you select:
    >
    > `This is an evaluation version.  There are [175] days left in the evaluation period.`
    >
-   > This message does not reflect the edition you selected. It relates to the preview period for RC1.
+   > This message does not reflect the edition you selected. It relates to the preview period for RC2.
 
 1. Once the configuration is done, verify that the service is running: