Archive

Archive for November, 2018

Azure SQL Managed Instance – Migrate a Database Using Backup and Restore

November 15, 2018 2 comments

If you have run through my last Managed Instance blog post, you have a Managed Instance at your disposal.  The PowerShell script for creating the network requirements also contains steps to create an Azure VM in a different subnet in the same VNet.  Unless you have a site-to-site VPN or Express Route between your on-prem environment and Azure, you will use this VM to connect to your Managed Instance.

Install Management Studio on the Azure VM.  To connect to your Managed Instance, you will need the host name for your Managed Instance.  You can find the Managed Instance host name on the resource page for your Managed Instance in the Portal.

GetMiHostName

Open Management Studio and copy the instance name into the Server field in Management Studio and enter the admin account and password you specified on creation.  Click connect.

ConnectToMi

Once the Managed Instance opens, you will be able to navigate it using Object Explorer.  The experience will look very familiar if you are used to using SSMS.  You will see all four system databases and SQL Server Agent.  If you expand the Agent entry in Object Explorer, you will notice that Alerts do not currently exist.  Also, note that the SQL version number is listed as 12.0.2000.8.  This is NOT SQL Server 2014.  Azure SQL Database and Azure SQL Managed Instance are currently at version 12.

ObjectExplorer

To create your first database, you will need to upload a backup file to Azure blob storage. If you do not have a Storage Account, click the Add button at the top of your Resource Group.

AddStorageAccount.png

In the search box, enter Storage Account.   Select Storage account - blob, file, table, queue.  On the next screen, click Create.

EnterStorageAccount.png

Confirm your subscription and resource group are correct. Select and name your storage account.  Select the same location as your Managed Instance, Standard performance, and Storage V2 (general purpose V2).  For this example, I selected Locally-redundant storage (LRS) because it is the least expensive option. Click the Review and Create button.

NameStorageAccount.png

Once the storage account is provisioned, you will need to create a container to hold the backup file.  I think of the storage account as a drive and containers as folders on that drive.  Under Services, click Blobs.

ClickBlobs.png

Once the Blob service screen opens, click Containers at the top of the screen.  Name your container and leave the Public access level set to the default.  Click OK.

ClickContainer

NameContainer.png

 

After your container is created, click the container name to open it.  Click the Upload button at the top of the screen.

UploadFile.png

Click the folder button to the right of the Files text box and select your backup file.  Once the Upload button becomes active, click it.  Once the upload is complete, you are ready to restore a database to your Managed Instance.

SelectFile

Go back to your VM and connect to your Managed Instance.  Right-click on Databases in Object Explorer and select Restore Database.

REstoreDatabase.png

Click the Device radio button.  Click the ellipsis button to the right of the Device text box.  Notice that URL is the only entry available in the Backup media type dropdown.  This is because you must restore a Managed Database from Azure blob storage.

SelectBackup.png

 

Click the Add button.  You will be prompted to sign into your Azure subscription.  Click the Sign In button.SignIn.png

Enter the login and password for your Azure account.  On the next dialogue, select your subscription, storage account, and container.  Click the Create Credential button to create a Shared Access Signature to allow SQL Server to access Azure storage.

SelectContainer.png

Once the credential populates in the Shared Access Signature Generated text box, click OK.

SelectCredential.png

You will return to the Select a Backup File Location dialog with the Azure storage container and Shared Access Signature boxes populated.  Click OK

OkBackupLocation.png

If you have used Management Studio to select a backup file to restore, the Locate Backup File in Microsoft Azure will look familiar.  Expand Containers in the left-hand pane and select your container.  You should see your backup file in the right-hand pane.  Click the file name and click OK.

SelectBackupFile.png

The Select backup devices dialog will display with your backup file selected.  Click OK.

BackupSelected.png

The Restore Database dialog will display with your backup file listed in the Backup sets to restore text box.  Note that there are no pages in the Select a page pane for Files or Options.  Because storage is managed for a Managed Instance, you cannot select where the files will be placed and you cannot set a database to NORECOVERY.  Click the Script button.

DatabaseRestore.png

A query window will open in Management Studio containing the RESTORE script that SQL Server will execute if you click the OK button on the Restore Database dialog.  Navigate to that script.  The RESTORE statement is extremely simple:

RestoreStatement.png

You cannot add any WITH clauses to the statement.  If you try you will receive the following error:

ErrorMessage

To complete the restore of your database, execute the restore script.  Once the restore completes, you will be able to access it using Management Studio. In my next post, I will provide additional options for migrating databases and instance-level objects to Managed Instance.

 

 

 

 

PASS Summit 2018 Day One Keynote

November 8, 2018 3 comments

"Any sufficiently advanced technology is indistinguishable from magic." -Arthur C. Clarke

In this morning's keynote session at PASS Summit 2018, public preview of a new feature in Azure SQL Database and SQL Server 2019 called Accelerated Database Recovery (ADR) was announced.  This changes the way that SQL Server handles recovery of a SQL Server instance on start up.

Historically, instance recovery runs three phases.  Phase one - Analysis - scans the transaction log for each database looking for two types of transactions:

  • Transactions that committed prior to instance shutdown but did not harden the associated data pages to the data file on disk
  • Transaction that started prior to instance shutdown without committed

Phase two - Redo - rolls forward the transactions identified in the first bullet point.

Phase three -Undo - rolls back uncommitted transactions identified in the second bullet point.

If long-running transactions need to be rolled forward or back, database recovery can take a long time.  In my experience, six hours is the longest recovery I have seen, but I know people who have seen much longer.

This leads me to Accelerated Database Recovery and the Arthur C. Clarke quote above.  ADR is not magic, and you can learn more about it at the link above, but it is a huge leap in SQL Server behavior and will pay huge dividends in database and instance availability.

I will continue my series on Managed Instance after returning from Summit, and will post more information about ADR once I get a chance to play with it.