A few idle thoughts…
Three months ago I had an issue regarding Windows Azure where I was unable to login to the management portal. At the time I contacted Azure support, the issue was soon resolved and I thought no more about it. Until today that is when I received an email from Azure support providing a detailed analysis of the root cause, the fix and moreover precise details about when and where things occurred.
The email itself is interesting and I have included the entirety of it below. A couple of things that leapt out at me:
- The level of detail and the diligence in investigating and subsequently reporting the issue back to me I found really rather impressive. They even outline the number of users that were affected (127 in case you can’t be bothered reading). Compare this to the quite pathetic support that another division within Microsoft, Skype, provided to Greg Low recently: Skype support and dead parrot sketches
“Windows Azure performed a planned change from using the Microsoft account service (formerly Windows Live ID) to the Azure Active Directory (AAD) as its primary authentication mechanism on August 24th. This change was made to enable future innovation in the area of authentication – particularly for organizationally owned identities, identity federation, stronger authentication methods and compliance certification. ”
I also found to be particularly interesting. I have long thought that one of the reasons Microsoft has proved to be such a money-making machine in the enterprise is because they provide the infrastructure and then upsell from that – and nothing is more infrastructural than Active Directory. It has struck me of late that they are trying to make the same play in the cloud by tying all their services into Azure Active Directory and here we see a clear indication of that by making AAD the authentication mechanism for anyone using Windows Azure. I get the feeling that we’re going to hear much much more about AAD in the future; isn’t it about time we could log on to
SQL Azure Windows Azure SQL Database without resorting to SQL authentication, for example? And why do Microsoft have two identity providers – Microsoft Account (aka Windows Live ID) and AAD – isn’t it about time those things were combined?
As I said, just some idle thoughts. Below is the transcript of the email if you are interested.
This is regarding the support request <redacted> where in you were not able to login into the windows azure management portal with live id.
We are providing you with the summary, root cause analysis and information about permanent fix:
Incident Title: You were unable to access Windows Azure Portal after Microsoft Account to Azure Active Directory account Migration.
Service Impacted: Management Portal
Incident Start Date and Time: 8/24/2012 4:30:00 PM
Date and Time Service was Restored: 10/17/2012 12:00:00 AM
Windows Azure performed a planned change from using the Microsoft account service (formerly Windows Live ID) to the Azure Active Directory (AAD) as its primary authentication mechanism on August 24th. This change was made to enable future innovation in the area of authentication – particularly for organizationally owned identities, identity federation, stronger authentication methods and compliance certification. While this migration was largely transparent to Windows Azure users, a small number of users whose sign-in names were part of a Windows Live Custom Domain were unable to login. This incompatibility was not discovered during the Quality Assurance testing phase prior to the migration.
Customers whose sign-in names were part of a Windows Live Custom Domain were unable to sign-in the Management Portal after ~4:00 p.m. PST on August 24th, 2012. We determined that the issue did impact at least 127 users in 98 of these Windows Live Custom Domains and had a maximum potential impact of 1,110 users in total.
The root cause of the issue was an incompatibility in the AAD authentication service to handle logins from Microsoft accounts whose sign-in names were part of a Windows Live Custom Domains. This issue was not discovered during the Quality Assurance testing phase prior to the migration from Microsoft Account (MSA) to AAD.
The issue was mitigated for the majority of affected users by 8:20 a.m. PST on August 25th, 2012 by running some internal scripts to correct many known Windows Live Custom Domains. The remaining affected domains fell into two categories:
Windows Live Custom Domains that were not corrected by 8/25/2012.
An additional 48 Windows Live Custom Domains were fixed in the weeks following the incident within 2 business days after the AAD team received an escalation from product support regarding those accounts.
Windows Live Custom domains that were also provisioned in Office365.
Some of the affected Windows Live Custom Domains had already been provisioned in AAD because their owners signed up for Office365 which is a service that also uses AAD. In these cases the Azure customers had to work around the issue by renaming their Microsoft Account or using a different Microsoft Account to administer their Azure subscription.
Permanent Fix: The Azure Active Directory team permanently fixed the issue for all customers on 10/17/2012 in an upgraded release of the AAD service.