Compartir a través de

Setting up SQL Server 2014 Always On Availability Groups with Azure Premium Storage using the new SQL Server 2014 Always On resource group template

  • Artículo

 

With Azure premium storage offering launched, I bet many of you have been wanting to use it for setting up their SQL Server 2014 Always on Availability Group clusters.

While, one can go ahead and set it up manually, what would be ideal is to use the available SQL Server 2014 Always On resource group template. A few clicks and you have the whole SQL Server Always On Availability Group environment setup.

At the time of writing of this blog, the template uses A3 Standard instances as default and a standard RA-GRS storage account. I am writing some steps below that you can implement to edit the setup created by the template for the SQL nodes to use Azure Premium Storage disks for Data & Logs.

These steps are based out of Keith’s blog on extending SQL Server Always On Availability Group setup, I have tweaked them to use Azure premium storage.

Please do ensure you follow all the steps and in the same order.

  1. Create the SQL Server Always On Availability Group setup from template with DS-series VMs

    The Azure premium storage account can at this point in time only be attached to the DS series VM. So our first step will be creating the template using DS series VM.

    Logon to the new portal at https://portal.azure.com and in the +New page click on Azure Marketplace. Type “SQL Server Always On” and hit enter you will find the template in the list (Figure 1).

    image
    Figure 1

    It loads the template, and you can enter a resource group name (make a note of the name as we will be creating a premium account in this resource group in next steps).

    The template uses A3 instances by default. Click on the SQL Server Settings option and click on Pricing Tier. It will load the recommended pricing tier, click View All and select the DS-Series VM you want to use. Please keep in mind, the VM instance SKU matters for the IOPS you need from the premium storage. The SKUs have the maximum IOPS number so it should be easy to identify the one that you need. I selected DS2 the for this which has 6400 IOPS (Figure 2).

    image 
    Figure 2

  2. Create premium storage account in the same Resource Group

    After some time elapses, while the servers are getting created we can go ahead and create a premium storage account in the same resource group. This is very important we want the account to be created as close to the VMs as possible. For doing this, please click on the +New button, Data + Storage and select Storage Account option.

    This will load the Create Storage account blade. Click on Pricing Tier and you will see list of storage account tiers. Please click “View All” and you will find the new Premium Locally Redundant option. Please select that and create the storage account. (Figure 3)

    image
    Figure 3

    Now let’s wait till the entire setup has been deployed. Once you get a confirmation notification on the dashboard.

    Please repeat the following steps for every SQL Node that has been created in the template.

  3. Disable Automatic Clustering

    Select the SQL Server node and then logon to the SQL Server using “Connect” option (Figure 4).
    image
    Figure 4

    Once you are connected, run the following powershell command to disable automatic clustering. We need to do this because we don’t want SQL Server to automatically cluster each time we add/remove a disk.

    Get-StorageSubSystem -FriendlyName "Clustered Storage Spaces*" | Set-StorageSubSystem -AutomaticClusteringEnabled $False

    I use powershell_ise.exe as I like the intellisense feature it provides. Once you have run this command, please keep the connection running and do not disconnect.

  4. Remove regular storage data disks and attach premium storage disks

    We need to remove the regular data disks that have been attached to the SQL Server node. To do that let’s go back to the portal, select the SQL server node, click on Settings which loads the settings blade and click on Disks.
    In the disks tab, select the data disks , right click and select Detach (Figure 6). Detach all the disks attached one by one (these are done sequentially so you will need to wait till previous one has been detached before you go to the next one).

    image
    Figure 6

    Once all the attached Data Disks have been detached, please click on “Attach New” disk. Click on “Select Storage Container” option, and select the new premium storage account you had created in step 2. Create a container “vhds” if you already haven't done that. Please make sure you select the premium storage account, as only this account tier has premium storage disks.

    Select host caching as “None” if you are going to use it for logging and select it as Read Only if its going to be used for data. (Figure 7)

    image
    Figure 7


  5. Setup disks for use

    Once you have attached all the disks, let’s go back to the remote desktop connection powershell window in the SQL Server and run the following powershell commands:

    $datadisks = Get-PhysicalDisk -CanPool $true

    This gets the list of data disks that have been attached and can be pooled.

    Note: If you plan to have both logging and data on the premium storage disks, please make sure you attached separate set of disks for each, and repeat 4 and 5 for each of them.

    New-StoragePool -FriendlyName "SQLDataPool" -StorageSubSystemFriendlyName "Clustered Storage Spaces*" -PhysicalDisks $datadisks | New-VirtualDisk -FriendlyName "SQLDataDisk" -UseMaximumSize -NumberOfColumns $datadisks.Count -ResiliencySettingName "Simple" -ProvisioningType Fixed -Interleave 65536 | Initialize-Disk -Confirm:$False -PassThru | New-Partition -DriveLetter J –UseMaximumSize

    This creates the drive J, please use any letter you like.  Once the drive is created, please run the following command to format it for use.

    Get-Partition -DriveLetter J | Format-Volume -FileSystem NTFS -NewFileSystemLabel "SQLDataVol" -AllocationUnitSize 65536 -Confirm:$False




  6. Change the default location for data and logging in SQL Server Node

    Please use the Link to change the data and log settings on SQL Server node.

You can run the SQLIO tool to get the details of IOPS and verify that you are maximizing the disk performance. After following these steps, I was getting 6450 IOPS.

Comments

  • Anonymous
    July 13, 2015
    SQL language is a very help fully language in this field

  • Anonymous
    July 14, 2015
    Nice guide

  • Anonymous
    July 14, 2015
    Thanks Kjetil!

  • Anonymous
    July 15, 2015
    Please STOP using wrong terminology. AlwaysOn is NOT a feature. It's a marketing term. Availability Groups is the feature. Azure gets this horribly wrong. MS is the worst abuser of this. Please correct your post and the portal also needs to be fixed. Horrible.

  • Anonymous
    July 15, 2015
    Thanks Allan for bringing this out. The intention was not to mislead but to use a term that everyone understands. I have edited the post to make it more clearer.

  • Anonymous
    August 31, 2015
    I was trying your procedure using GS Vms but I am not able to add premium storage after detaching the data disks, any ideas why?  

  • Anonymous
    October 07, 2015
    I'm having the same issue as user 'none'. The premium storage account isn't listed in the list of existing storage accounts (after detaching the original disks and attempted to attach a new disk).

  • Anonymous
    October 07, 2015
    I've solved the issue. You must deploy the storage account using the 'classic' model. In the preview portal, don't choose 'Storage Accounts' from the left nav. You have to choose New -> Data + Storage -> Storage Account. Classic is the default. If you use the left nav and click the 'New' button, the storage account will be created using the Resource Manager deployment model, which doesn't work.

  • Anonymous
    October 20, 2015
    Thanks Brian for pointing this out! Yes, you need to select the default Storage account that comes with the template to start with. A point I missed highlighting. Apologies, but it's been a while I monitored this post and didn't see your questions.