Thanks,

Francesco

@francedit

Ciao a tutti,

vi segnalo questo incontro dove affronteremo la tematica dei Big Data. Di seguito potete trovare i dettagli.

Quando: 27 novembre

Dove: Microsoft Innovation Campus, Peschiera Borromeo(MI)

Registrazione: https://msevents.microsoft.com/cui/EventDetail.aspx?EventID=1032603304&culture=it-IT

Francesco

@francedit

I will speak at WPC2014 in Assago in November. I will talk about Cloud Computing (Azure), and SQL Server, focusing on the new features available, like SQL Database new tiers, Machine Learning, and other Azure cool new functionalities! Details on my sessions (and quantity of sessions that I will deliver :)) are not 100% ready yet, but you could visit the conference agenda to get more details.

See you in Assago(MI) on November 18^th,19^th,20^th  !

Francesco

@francedit

This blog post is composed of two parts:

In the first part, we will see how to load data from Azure SQL Database using the R Language installed on my local machine.

In the second part, we will access the same data using Azure Machine Learning, a fully managed cloud service for data scientists and developers, currently in preview, which provides a great support to language R.

Prerequisites

To load data from Azure, we need first to create an Azure SQL Database. If you are new to Azure, you can activate a 30 days free trial here. The process of creating a new SQL Database server is very easy, and a getting started guide is available here. Once created, you will have access to a control panel, represented in the image below.

SQL Database is a fully managed service, and Azure will take care of things like high availability and backups.

To design the database schema and load data, you can use SQL Server tools like SSMS, which are fully compatible. In our example we will use a dataset with the history of winners of the Italian soccer league (I am a Juventus supporter, so I like the query results shown below very much :) )

Part 1 – load Azure SQL Database data using R

In the first part, we will use R installed locally on my Windows 8.1 laptop. To accomplish this part we need to:

1. Configure an ODBC dsn to Azure SQL Database

2. Download and Install R bits here: https://cran.r-project.org/bin/windows/base/

How to create an ODBC dsn

In the first step of the wizard, we need to provide a name for the dsn (sqldb) and the name of the Azure SQL Database Server ([servername].database.windows.net)

We need to specify the default database (ItalianSoccerLeague)

And SQL login and password

Using R to load SQL Database data

Now we have our ODBC dsn ready. R has lots of packages available, and one of them, called RODBC allows the communication with ODBC data sources. To install a package R we can use the command install.packages(“[package name”)]. The package will be downloaded from the web

Once that the package is installed, we need to load the package, using the following sintax

Now we can open a connection to the Azure SQL Database, using the ODBC dsn, plus the User ID and the Password. odbcConnect is the command that we will use, and the result is assigned to the sqldbcnn variable

Last step, using sqlQuery, is to execute the query

If we need to get a summary of the results, we can use the summary function, passing the results variable as a parameter.

Part 2 – load Azure SQL Database data using Azure Machine Learning

Azure Machine Learning runs inside Microsoft Azure, and it is currently in preview. To get started with Azure Machine Learning you could visit the following page: https://azure.microsoft.com/en-us/trial/get-started-machine-learning/

As we have done in part 1 with R running on my laptop, we will now use Azure Machine Learning to load data from Azure SQL Database. Azure ML provides a designer, called Azure ML Studio, which is a workbench that runs in the browser, e.g. Internet Explorer. You don't need to install anything on your local machine, everything runs inside Azure

First thing to do is create a new experiment; this will give us access to a graphical designer, the place where the data scientist will spend most of his time.

Azure ML is made of many modules, some of them can helps us to load and manipulate data. In this case, we need the Reader module, and we just need to drag and drop it into the designer.

Each module has a properties pane, where we can set all the parameters required to connect to Azure SQL Database, including the query string.

That is all we need to do to get data from Azure SQL Database. As an optional step, we could also use a Descriptive Statistics module, and connect it to the Reader module. This will help to have a more detailed view of our dataset

Now we can run our experiment using the RUN icon  

We can visualize the data received from the database using the Visualize command in the context menu.

In addition, we could get some additional descriptive information, like a MultiboxPlot to see a graphical representation of our data, or Mean, Min and Max values.

Francesco

@francedit

[this post is written in italian as the workshop will be delivered in italian]

Big Data, In-memory e Cloud sono alcuni dei trend da cui le aziende possono trarre vantaggio oggi per innovare le proprie realtà, i propri processi e ottenere nuovi vantaggi competitivi. Grazie alle più recenti tecnologie rilasciate da Microsoft come SQL Server 2014 e Power BI per Office 365 aziende di qualsiasi dimensione possono iniziare a implementare progetti di Big Data, Self-Service BI e a migliorare i livelli di disponibilità e le performance delle applicazioni in modo semplice, efficace e a costi contenuti.

Microsoft e UGISS ti invitano a questo evento gratuito in cui potrai scoprire, tramite demo live fatte da riconosciuti esperti del settore, come la tua azienda potrà trarre beneficio immediato da tecnologie quali In-memory, HDInsight, Power BI e dalla piattaforma Microsoft Azure.

Agenda

Registrazione

La partecipazione all’evento è completamente gratuita e per partecipare è necessaria la registrazione

Informazioni logistiche

When I visit customers, the curiosity about Azure is always very high. One of the most frequent questions I get is:

“Can you help me to identify what I can do to modernize my datacenter leveraging some of the hybrid services that Azure offers?”

Well, that is an interesting question, but a simple blog post is certainly not the right place to answer it entirely. However, we can try to focus on a specific scenario here, so let me rephrase the sentence in this way:

“I already have SQL Server installed in my datacenter; I would like to utilize Azure to do backups, or to design a High Availability / Disaster Recovery solution. What can I do with my existing on-premises environment?”

To answer the question I decided to build a simple table that includes the most commonly used features that SQL Server provides for backups and HA/DR scenarios, like Log Shipping, DB Mirroring and AlwaysOn Availability Groups.

I am not covering SQL Server Failover Cluster here, as it is not a supported scenario in a hybrid configuration with Azure. Furthermore, I am not covering external tools; I am just focusing on SQL Server embedded features or, in case of backups, tools shipped by the product group to allow older versions of SQL Server to backup files to Azure Storage.

The table below assumes that the customer has a SQL Server installation already in-place, and he needs to extend the on-premises datacenter to use Azure for backups or HA/DR. First column is the version of SQL Server, and I have started from SQL Server 2005 up to SQL Server 2014. The other columns represent the features that we just mentioned above. I hope it will help you with the conversations with your customers.

*requires a VPN connection between the on-premises network and Microsoft Azure

** Synchronous commit could affect performances in hybrid configurations

*** Database Mirroring is a deprecated feature

Further Information

SQL Server 2014 enables the possibility to store database files as Azure Blobs. You can learn more about this feature here, or read this Whitepaper

Francesco

@francedit

SQL Server 2014 and Microsoft Azure Blob Storage service: Better Together

Francesco

@francedit

 You can download slides and demos using links below:

DEMO : https://francedpersonal.blob.core.windows.net/pub/DEMO_cdays14.zip

Francesco

@francedit

I have already published a couple of articles talking about Windows Azure and SQL Server in hybrid scenarios. I will reference them later in this document as they explain in detail some configuration step needed by this post and for which I won’t go to much in detail here.

First thing to underline is that the Add Azure Replica wizard is a client tool, so we just need to install SQL Server 2014 Client Tools to use it. It targets SQL Server 2014 (now in CTP2) but also SQL Server 2012 SP1. Supporting SQL Server 2012 is interesting for those customers whom have just finished upgrading and, at the same time, they want to take advantage of the cloud features for disaster recovery, backup or resource offloading purposes.

Prerequisites for the Add Azure Replica wizard

1. An AlwaysOn Availability Group configuration already in-place in the on-premises environment.

2. A Windows Azure subscription that will host SQL Server AlwaysOn secondary replicas

3. A VPN Site-To-Site between your on-premises network and Windows Azure.

Test environment topology

The left side of the image below shows the on-premises environment (part of the prerequisites). The Add Azure Replica wizard will configure the third node (the right side of the image); in my example, it is the 10.0.0.0 Virtual Network. In other words, SQL Server will take care of this step for you! We just need to have the Virtual Network and the VPN tunnel properly configured.

ON-PREMISES Virtual Machines and installed Features/Roles

Below you can find the configuration that I have used for testing.

- SQL2014DC (IP 192.168.0.18). Domain Controller (in my example I have created an Active Directory domain called dotnetisland.org), DNS Server (also used by the Windows Azure Virtual Network for name resolution), SQL Server Management Tools

- SQL2014N1 (IP 192.168.0.16). Member Server, Windows Server Failover Cluster (e.g. cluster1.dotnetisland.org), AlwaysOn node 1 (synchronous replica), user databases

- SQL2014N2 (IP 192.168.0.19). Member Server, Windows Server Failover Cluster (e.g. cluster1.dotnetisland.org), AlwaysOn node 2 (synchronous replica), user databases

In order to mimic a real environment, it has been created different accounts to separately execute services and to run the Wizard. It has been configured a service account for SQL Server (sqlservice) and a user account to run the wizard (sqluser). Secondarily, it has been added two databases to the AlwaysOn group, and each database has database and log files on different disks. The wizard does a great job in recognize the on-premises setup and applying the appropriate configuration to the remote node.

Windows Azure network configuration

It has been configured a site-to-site VPN and specified the on-premises DNS server. To create a site-to-site VPN you need to either configure a VPN hardware device, or use Windows Server 2012 with the RRAS service. Details on how to configure a site-to-site VPN are available at this link or you can visit my previous blog post. There is also a good tutorial available on the Windows Azure website, here. This step could be quite complex for a DBA who is not also a network engineer. Therefore, especially in configurations that require the configuration of a VPN hardware device (e.g. a Cisco or a Juniper firewall), I recommend to ask the help of a network administrator. Once the VPN will be up and running, you will see Windows Azure as an extension of your datacenter, and the rest of the configuration is straightforward.

Please note that Windows Azure Virtual Networks do not allow Static IP Addresses. When deploying a Virtual Machine in Windows Azure, a DHCP Server will assign an IP address from the subnet you specify. It will stay with the VM for its lifetime, unless you decide to stop and deallocate it.

The image below, taken from the Windows Azure portal, describes the site-to-site VPN configuration that I have used.

francedsitetosite represents the Virtual Network name

MyOnPrem is the on-premises subnet

When you see that the half-blue and half-green bar connects the two blocks (in my case francedsitetosite and MyOnPrem), it means that the VPN tunnel is working fine.

DNS Servers block contains the list of the on-premises DNS servers used for name resolution. You can set more than one DNS Server if you want. In my example the DNS Server is also the Domain Controller for the Active Directory domain called dotnetisland.org

DATA IN and DATA OUT indicate the amount of data moved in and out the VPN tunnel.

GATEWAY IP ADDRESS (hidden by the black bar in the image) is the IP Address of the Windows Azure Virtual Gateway that will establish the VPN tunnel

Adding a third Windows Azure node using the Add Azure Replica wizard

The Wizard can be started from the “Add Replica…” context menu in SSMS. We will add a third, asynchronous replica hosted by a Virtual Machine that we would call e.g. FDIAZSQLNODE3 

First thing to do is to connect to your Windows Azure subscription. In order to perform this task, the wizard allows the connection to the Windows Azure portal to download a management certificate.

After the connection to the Windows Azure portal has been established, we have to specify the following parameters:

• Image type (in my example SQL Server 2014 CTP2 Evaluation on Windows Server 2012 R2). The available SQL Server images from the Image Gallery will fill the dropdown list.
• Image size (a Large VM is the minimum recommendation for a production environment)
• Virtual Network name and Virtual Network Subnet name
• Virtual Machine name and Virtual Machine local credentials
• Domain name and Domain user credentials

And this is it…

Following screenshots are the same as adding an AlwaysOn replica on-premises. Here, for example, we could decide to set the secondary as readable.

The screenshot below provides also a link to check the report log, that can help us to troubleshoot in case of errors.

What the wizard has done for us (behind the scenes)?

• It has created a Windows Azure Storage account

• It has created a cloud service and the Windows Azure VM will be assigned to it. If you will add additional AlwaysOn replicas in Windows Azure, they will share the same cloud service.

• The name of the cloud service will be: alwayson+[name of the Virtual Network]+[name of the Availability Group]+GUID

• It has provisioned the virtual machine, using the image of our choice, and consequently added the data disks needed by SQL Server

• The virtual machine has been added to the Active Directory domain (in my example the domain is dotnetisland.org)
• The Failover Cluster feature has been installed and the VM has been joined to the cluster
• Configured the AlwaysOn availability group and the databases.

In case of failures (e.g. connectivity issues), a detailed log is available and the steps already taken will be rolled back.

Considerations and resources

Several tests, using database of small size, took less than 40 minutes to complete, and considering that it has not been configured anything on the Azure side (except VPN), this might definitively be a feature that DBAs should explore.

I have also tried to challenge the wizard, adding more nodes to the cluster, simulating several failures and I can say that, even if I am using a CTP version, it is already stable and it works very well. Below you could find some links with additional information.

https://msdn.microsoft.com/en-us/library/dn463980(v=sql.120).aspx

https://msdn.microsoft.com/en-us/library/ff877884(v=sql.120).aspx

https://blogs.technet.com/b/francesco_diaz/archive/2013/11/15/sql-server-2014-ctp2-alwayson-availability-groups-hybrid-it-scenario-using-windows-azure.aspx

Francesco

@francedit