As far as Azure IaaS goes, this is the biggest improvement to the platform since the ExpressRoute offering.

The announcement is here, and I highly recommend reading it:
http://azure.microsoft.com/blog/2015/03/26/azure-backup-announcing-support-for-backup-of-azure-iaas-vms/

The highlights:

• With Azure Backup, you can now get application consistent backup of Windows VMs without having to shut down the VM.
• “In order to backup IaaS VMs, the customer needs to deploy absolutely nothing”*
  Note: This is accurate insofar as you have the Azure VM Agent installed (see Prerequisites below)
• Azure Backup truly achieves “set-and-forget” for VM backups.
• Azure Backup does additional processing to determine the incremental changes between the last recovery point and the current VM state. By transferring and storing only the incremental changes, Azure Backup is highly storage efficient.

Azure VM Agent Prerequisite

• A very important prerequisite is that the Azure VM Agent must be installed. This is performed when the VM is first created, but if you uncheck the box to install the agent, then you will not be able to back it up with the new VM level backup feature.
  
• If you do not have the Azure VM Agent installed, you will get an error message during the registration job step:
  ”Failed to install the Azure Recovery Services extension on the selected item. VM Agent is a pre-requisite for Azure Recovery Services Extension. Please install the Azure VM agent and restart the registration operation.”
• You can manually download and install the VM Agent if it is not installed on the VM, see this article for more information:
  https://msdn.microsoft.com/en-us/library/dn832621.aspx
• The VM agent itself can be downloaded directly from (here) and is very small and takes seconds to install.
• After manually installing the agent, it is necessary to set the ProvisionGuestAgent value to true using Powershell or a REST call. If you do not set this value after manually installing the VM Agent, the addition of the VM Agent is not detected properly.)
• See my blog post for manually installing the VM agent for step by step instructions:
  http://tctblgs.azurewebsites.net/manually-install-the-azure-vm-agent/

Seeing it in Action!

Assuming you have already setup your recovery vault, and your VM’s have the VM Agent installed, then there are three easy steps to start backing up VM’s in Azure IaaS

After clicking on ‘Discover Virtual Machines’ you then click on the Discover button at the bottom of the screen.

After discovery completes, you then click on the Register button.

This brings you to a screen to select the VM’s that you want to protect.

Clicking on the checkmark will spawn a register job that can be viewed on the Jobs tab. In my case, this job took 4 minutes to run.

Now that we have a VM registered, the next step to perform is step number 3, “Protect Registered Azure Virtual Machines.

Select the item you want to protect

You can then select an existing policy or create a new policy

When the backup has taken place, you can view it under protected items.

You can force a new backup or you can click on Restore to bring the VM back to life as a new VM standing next to the old one (it does not overwrite the existing VM).

Restoring a backup

A backup is only “good” if you can verify it by performing a restore. Until then, you should not trust your backups. I have learned this the hard way in the trenches =)

It is interesting that when you restore a VM, it does not overwrite the existing VM, but it instead deploys it alongside the current VM.

You can check the Jobs tab to see how long the restore will take. In my case, the restore took 23 minutes.

When the restore job completes, you can view the job notes:

To view the restored VM, I had to sign out and back into the Azure Management portal, but then I saw the restored VM amongst my other VM’s:

This raises a practical operational question where you need to be sure to shut off the old one before the new one starts up otherwise in a batch environment you wouldn’t want two VM’s running the same batch (you get the idea, you need to know what your VM’s are doing and coordinate properly).

Therefore, it would be good to have an option during the restore process for Azure to automatically shut down the original VM on your behalf to tighten-up the handoff, as you want to avoid having two machines with the same computer name and SID running on the network.

For example, in my restored VM, I can see it still has the original computer name (as I would expect) and so even though the name of the VM in Azure shows as ‘MyRestoredVM’ the actual computer name maintains the original name. (This is okay behavior, but just remember we need to shut off the original VM now too). I posted this feedback on the Azure Feedback portal, please click (here) to vote if you agree and would like the Azure Product team to include this feature in a future release.

When you first deploy a Virtual Machine to Azure using the Gallery, you have the option of installing the VM Agent. You should always leave this box selected because it adds tremendous value.

VM extensions can help you:

• Modify security and identity features, such as resetting account values and using antimalware
• Start, stop, or configure monitoring and diagnostics
• Reset or install connectivity features, such as RDP and SSH
• Diagnose, monitor, and manage your VMs
• Backup your VMs with the new Azure VM level Backup feature (see my blog post here for more information on that feature announced March 26, 2015).

The agent allows you to add extensions to the VM, for more information, see this article:
https://msdn.microsoft.com/en-us/library/dn832621.aspx

You can manually download and install the VM Agent if it is not installed on the VM, see this article for more information:
https://msdn.microsoft.com/en-us/library/dn832621.aspx

The VM agent itself can be downloaded directly from (here) and is very small and takes seconds to install.

After manually installing the agent, it is necessary to set the ProvisionGuestAgent value to true using Powershell or a REST call. If you do not set this value after manually installing the VM Agent, the addition of the VM Agent is not detected properly.) The following code example shows how to do this using PowerShell where the $svc and $name arguments have already been determined

$vm = Get-AzureVM –serviceName $svc –Name $name
$vm.VM.ProvisionGuestAgent = $TRUE
Update-AzureVM –Name $name –VM $vm.VM –ServiceName $svc

Before you can run the powershell commands above, you first need to install the Azure Active Directory powershell module and then run the add-azureaccount command. Then to make sure you can see your VM’s, you can run get-azurevm.

In this screenshot you can see I have two VM’s, one with the agent installed and one without it.

So after manually installing the VM Agent, I can then run these commands to update Azure so that it knows the agent has been installed inside the guest VM:

$vm = Get-AzureVM –serviceName TCTCloudService –Name MyBackupTest
$vm.VM.ProvisionGuestAgent = $TRUE
Update-AzureVM –Name MyBackupTest –VM $vm.VM –ServiceName TCTCloudService

Now when I check for guest agent status, I can now see the same for both VM’s:

In this blog post I am going to review the upgrade process of Dirsync to the new AAD-Connect. The March 2015 preview now makes it possible to perform an in-place upgrade from Dirsync to AAD-Connect. This entire process took 30 minutes for me in my lab environment, but your performance and time may vary because I am running a small environment on SSD hard disks =).

Important: You must read the “Azure AD Connect Public Preview 2 Readme” file – there are too many requirements and prerequisites in that Readme file to summarize in this blog post, so please do not skip that reading.

I also recommend reading the “New Sync features in Azure AD Connect Public Preview 2.docx”

You can download the AAD-Connect March Preview, Readme file, and the New Sync Features document from the Connect site here:
https://connect.microsoft.com/site1164/Downloads/DownloadDetails.aspx?DownloadID=53949

After the prerequisites are installed, AAD-Connect detects Dirsync and will now upgrade it in-place:

Next, I am prompted to enter my Azure AD Administrator Credentials

Lastly, I am ready to click Install. I recommend unchecking the box to start synchronizing after install. You can always start it manually later when you are ready.

This part of the wizard took about 10 minutes to uninstall Dirsync and install AAD-Connect.

Next, I clicked on the icon on the desktop “Azure AD Connect”

I then signed in.

I then type in an Enterprise Admin account

If I want to connect to an additional forest, I can do that here:

Next, I select that I want to enable password writeback. You will notice that user, group and device writeback are greyed out and not selectable. This is because we have not yet run the AD preparation steps necessary to enable those features. See the bottom of this blog post for details on enabling those features.

And one last confirmation by clicking Install

And one last confirmation that the installation was successful

Enabling User, Group and Device Writeback

In the readme file, it describes the powershell commands to run that will enable this enhanced functionality.

For group writeback, your on-prem Exchange server must be running Exchange 2013 CU8. Also, the default Sync Rules will not add the address book attribute. Find the value from your Exchange server and add this as a custom attribute flow.

The Initialize-ADSyncDomainJoinedComputerSync Function will initialize your Active Directory forest to sync Windows 10 domain joined computers to Azure AD. This function will need to be run on each forest to allow Windows 10 computers to authenticate against ADRS.

The Initialize-ADSyncDeviceWriteBack Function will initialize your Active Directory forest for write-back of device objects from Azure AD to your Active Directory. It will also set up the necessary permissions on the AD Connector account. This only needs to be run on one forest even if AzureADConnect is being installed on multiple forests.

Note: I received errors when running this command.

The Initialize-ADSyncGroupWriteBack Function will initialize your Active Directory forest for write-back of group objects from Azure AD to your Active Directory. It will grant permissions to an AD Connector account for modifying objects in a pre-existing group WirteBack container. Please use this same container for Group WriteBack when you run the wizard. This function only needs to be run in one forest.

I created a new organizational unit for these objects called “CloudUsersAndGroups”

Initialize-ADSyncGroupWriteBack -GroupWriteBackContainerDN “OU=CloudUsersAndGroups,DC=thecloudtechnologist,DC=com”

The Initialize-ADSyncUserWriteBack Function will initialize your Active Directory forest for write-back of user objects from Azure AD to your Active Directory. The users will be created with a random password so you have to reset the password in ADDS for the user to actually be able to login.

Initialize-ADSyncUserWriteBack –UserWriteBackContainerDN “OU=CloudUsersAndGroups,DC=thecloudtechnologist,DC=com”

Note: The Azure AD Premium feature password writeback does not work for users configured for user writeback. In other words, if you have a cloud identity, and that user is synced to the on-premises AD, then the password writeback feature will not update the newly created on-prem AD account version of the cloud identity user. I assume it would still reset the cloud identity portion.

After running these commands, I went back to the wizard but the options were still greyed out. This may be because my AD Schema is not running Exchange 2013 CU8, so I will update my schema and then update this blog post after that is done.

Next, read how to configure Azure AD Password Write-back on MSDN (I recommend reading all seven (7) articles under ‘Password Management’
https://msdn.microsoft.com/en-us/library/azure/dn510386.aspx 

In the Azure AD Tenant, I enabled the toggle “USERS ENABLED FOR PASSWORD RESET”

And when I scroll down, I now see that Password write back service status is ‘Configured’

What does the user experience look like for self service password reset?

Typically, the user will click on the “Can’t access your account” link below the Office 365 sign-in page at http://portal.office.com

Otherwise, they can also bookmark directly to the self-service password reset page:
https://passwordreset.microsoftonline.com

They will be prompted to authenticate with text message, email or phone call. You can configure which of these options you want the user to enter. The user can also register for self service password reset and populate this contact information in advance, or an administrator can pre-populate it (again, please read the MSDN articles above for more details).

The user can then select the new password which must conform to the on-premises password policy.

This is a follow-on post from my post on Azure Application Proxy. Assuming that you have published your first application via the Azure Application proxy, you may now want to secure it with Multifactor authentication.

You can enable access rules to limit access to the application you publish with Application Proxy to specific groups, you can require multi-factor authentication, or only require MFA when the user is outside a specific network location (external IP address of NAT firewall).

The first time MFA is enabled for an application published by Azure Application proxy, the user will be required to enroll in MFA.

After enrolling, the user will be sent a text message or a phone call to their phone number registered in AD. Then they will logon to the application with two forms of authentication (password + phone call or text message).

The next time they browse to the application, after authenticating with their username and password, the application will automatically send them a text message and they can then sign in after entering the SMS code sent to their smart phone.

The intranet application hosted internally at https://intranet will then load up fine.

Azure AD Application Proxy (AAD-AP) is a type of reverse proxy solution that enables access to web-based applications that exist on a corporate LAN, secured behind a corporate firewall.

The benefits of using AAD-AP rather than using a traditional firewall to expose an application to external access are (1) the convenience of listing the application in the user’s Office 365 menu choices (see first screen shot below) or the Azure Access panel and (2) the enhanced security of preauthentication using Azure Active Directory and the option of enhancing security with Azure Multifactor Authentication. These later two security enhancements were previously provided by solutions such as Microsoft ISA or TMG.

The convenience that users benefit from is having one place to access all internal web-based corporate applications, as well as over 2,400 3rd party SaaS applications. In the screen shot below you will see that amongst the Office 365 applications list, I have also configured single sign on for Facebook, Google Docs, ADP, Salesforce, and more. It is very convenient to logon once to Azure Active Directory or Office 365, then launch other applications without having to logon to those applications individually.  This amounts to huge time savings and it is really nice not having to remember 10 separate usernames and passwords! This now puts Azure AD on par with other hosted identity providers such as Okta, Onelogin or PingFederate.

This blog post is a review of AAD-AP, a component of Azure AD Premium and Azure AD Basic.

AAD-AP exposed application ‘My Intranet’ in Office 365

If you don’t have Office 365 you can also use the Microsoft Azure access panel to achieve SSO (as shown below).

http://myapps.microsoft.com (this redirects you to https://account.activedirectory.windowsazure.com/profile/ )

The 3rd way of accessing internal applications through AAD-AP is a direct hyperlink. This is provided after configuring the application in the Azure management portal.
For example: https://myintranet-tctechnologist.msappproxy.net/ 
The convention is: (Application Name – Azure Tenant Name – MSAppProxy.net (similar in concept to the O365 tenant name “onmicrosoft.com”)
Custom domain names are coming soon, so you will be able to have the AAD-AP name in front of your own domain name, ex: https://myintranet.contoso.com 

Application Proxy Prerequisites

• You must have an Microsoft Azure administrator account. If you don’t’ have one, you can get a 30 day trial here.
• You must have an Azure AD Premium license. For more information, see Getting Started with Azure AD Premium. You can also get a 30 day trial to evaluate this as well.
• A server running Windows Server 2012 R2 or Windows 8.1 or higher on which you can install the Application Proxy Connector. The server must be able to send HTTPS requests to the Application Proxy services in the cloud, and it must have an HTTPS connection to the applications that you intend to publish.
• The server running the connector must be able to make outbound connections to this domain and subdomain: msappproxy.net on the following TCP/IP port ranges:
  443, 20200-20210, 9352, 10100-10120, 8080, 9090 and 9091. For a description of what these ports are used for, see : https://msdn.microsoft.com/en-us/library/dn768214.aspx
  • There are no inbound ports required, because Azure Application Proxy service (ApplicationProxyConnectorService.exe)  initiates a reverse tunnel from the VM out to Azure.
• If the web application requires windows integrated authentication, then the machine where the connector is installed must be joined to the domain.
• For windows integrated authentication, The UPNs in Azure Active Directory must be identical to the UPNs in your on-premises Active Directory in order for preauthentication to work. Make sure your Azure Active Directory is synchronized with your on-premises Active Directory.
• For accessing applications remotely, the application that you are proxying cannot send the user any 302 redirects otherwise those will most likely contain internal server names that are not accessible internally. See the scenario below that happens when I first tried to publish the Lync control panel. The fix for the Lync scenario was to publish the actual path to the CSCP virtual directory. This is thanks to the new Path Publishing feature that was announced on March 11th, 2015 on the Azure Application Proxy blog here.
• In my testing, AAD-AP worked with Windows machines running Internet Explorer or Chrome, iOS devices and Mac OSX (Safari and Chrome). The official support statement says that the Access Panel Extension is available for Internet Explorer 8 and later, Chrome, and Firefox browsers.

Microsoft AAD Application Proxy Connector

The connector is a small 4 MB file that can be downloaded from the Azure Management Portal when configuring a new application that relies on Application Proxy.

You should install the connector software on a VM that has HTTP/S access to the application that you want to publish, and outbound access to Azure.

Installing the connector is quick and painless. The only information you are prompted to provide is just the Azure tenant administrator username and password.

It installs two services, the main service runs under network service because it must be able to interact with other servers to perform the reverse proxy, and optionally to perform Kerberos delegation for web based applications that required windows authentication.

Configuring the Connector for Windows Integrated Authentication

Let’s say you want to publish the Lync Control Panel to through the Azure application proxy.

Since the Lync Control Panel requires Windows Integrated Authentication, we need to configure the Active Directory Computer object for delegation. For step by step details, see this Microsoft article, otherwise I will cover the high level steps here.

I did not have a service principal name (SPN) for my Lync simple URL for administration, I had to create it first. Before creating an SPN, it is good practice to check to see if it exists first. You can query to see if an SPN exists with the ‘setspn’ command that you run in a regular command shell on any server (domain admin privs are required).
setspn -Q http/admin.thecloudtechnologist.com
Note: The SPN format is a little odd because there is a single forward slash, whereas you would normally expect to see a colon followed by two slashes.

To create the SPN, I ran this command:
setspn –S http/admin.thecloudtechnologist.com tctfe01  
(Note: The Lync ‘standard edition’ front-end server name was TCTFE01. For Enterprise editions, you will use a service account rather than an individual computer name). You can use the –Q or –L parameters to verify that it registered correctly.

After the SPN has been created, the next step is to configure the Active Directory Computer object for delegation. Find the computer object for the machine where you installed the Connector software in Active Directory Users and Computers. In my case, my connector was installed on a server named “HV1” so I found that object and on the delegation tab, I added the computer name of my Lync standard edition front end server “TCTFE01.” For Enterprise Edition deployments, you would configure delegation on the relevant service account instead.

This allowed me to find the SPN I had created for the SPN record admin.thecloudtechnologist.com

Now you can add the SPN into the Azure Management portal:

You should now see the Lync Control panel in the Office 365 list of applications.

What I found out next helped me to understand more about the features and limitations of Azure Application Proxy.

When I published the simple URL for the Lync control panel (‘admin.thecloudtechnologist.com’) I found that it was working internally but it failed to work externally. After some fiddler analysis I discovered that it was working internally because Lync web services sent an HTTP 302 Redirect to the internal DNS host name of the Lync standard edition front-end server, along with the virtual directory of the control panel, ex: (‘tctfe01.thecloudtechnologist.com/cscp’). Since the internal host name is not published in external DNS, nor is there a firewall rule permitting this, it failed externally.
So Azure Application Services is faithfully sending the 302 redirect back to the external user, so there was nothing broken, per se. This shows one requirement to be aware of – check to make sure that the application you want to proxy does not use 302 redirects.

HTTP/1.1 302 Redirect
Content-Length: 168
Content-Type: text/html; charset=UTF-8
Location: https://tctfe01.thecloudtechnologist.com/Cscp
Server: Microsoft-IIS/8.0 Microsoft-HTTPAPI/2.0
X-Powered-By: ASP.NET

<head><title>Document Moved</title></head>
<body><h1>Object Moved</h1>This document may be found <a HREF=”https://tctfe01.thecloudtechnologist.com/Cscp”>here</a></body>

The workaround that I found was to use the new path publishing feature (Thanks Microsoft Application Proxy team!!). This enabled me to publish the full path where the 302 redirect was sending the user to (‘https://tctfe01.thecloudtechnologist.com/Cscp’). This can only be setup when the application is first configured, so if you forgot to add /cscp in the beginning, you will need to delete the app and start over.

In my case, I didn’t have /cscp/ from the beginning, so I had to delete my app and start over like this:

There was already an existing SPN record, so I did not have to create another SPN for tctfe01.thecloudtechnologist.com, but I did have to configure computer delegation on the HV1 computer account for http/tctfe01.thecloudtechnologist.com instead of http/admin.thecloudtechnologist.com.

So even though the internal dns name of my Lync server is not published externally, this shows the power and benefit of Azure Application Proxy because I can now access the Lync administrative control panel from anywhere.

Recommendations

1. The connector software is not highly available, so make sure to install it on a virtual machine that benefits from high availability at the hypervisor layer (ex: vmotion or live migration).

2. Since this is such a new service, I am not sure what the scalability and performance of the connector service would be in a production environment. So you would need to perform your own performance testing. Consider using Azure’s Visual Studio Cloud Load Testing. 

3. In the next blog post in this series, I describe the new conditional access feature using Azure’s multi-factor authentication. See Controlling Access to Application Proxy (Optional)