"The cloud is down." Words you never want to utter to your CEO.
Every major cloud provider has experienced outages: Amazon, Google, and most recently, Azure.
It looks like Microsoft tried to roll out a single patch across its global infrastructure and everything failed at once. This understandably annoyed customers who paid extra for geographic redundancy. It also led to the usual concerns about the paucity of redress and compensation in the Azure SLA. At least Microsoft -- unlike Salesforce -- gives you an SLA in the first place.
The error was compounded by poor customer communications when the Azure availability dashboard displayed outdated information, since the dashboard itself was running in....Azure. Redmond turned to Twitter instead, which was clever, but also rather confusing to customers.
[This reminds me of an old joke among SharePoint insiders:
- "Where did the junior system administrator store his SharePoint disaster recovery documentation?"
- "I dunno, where?"
- "In the DR document library within IT's SharePoint site!"
Hah, hah, hah, hah...]
What This Means for SharePoint and Yammer Customers
If you're running SharePoint Online (part of Office 365) or SharePoint on-premise, ostensibly you were shielded from direct impact. O365 gets served to you from in its own, separate data centers that are not part of Azure. And Yammer runs within yet a separate set of infrastructure. Get more details on this from RSG's social-collaboration vendor evaluation research.
So all good, right?
With SharePoint 2013, Redmond has largely abandoned previous extension approaches in favor of the "app model," which encourages you to build atomic services above SharePoint, rather than within SharePoint. Then Microsoft encourages you to put those apps in Azure. So with an Azure outage, your SharePoint system may not go down, but parts of it could get broken.
Also, if you want to enjoy single sign-on across on-premise systems and Office 365, Microsoft will point you to Active Directory Federation Services (AD FS), and again encourage you to host those services in Azure. For some customers, this meant the Azure outage locked them out of O365.
Redmond sees Azure as the glue that puts its various off-premise pieces together. In fact, Azure is becoming especially important for SharePoint in the cloud. As subscribers to our SharePoint research know, Redmond has developed a co-existence model, but, astoundingly, never developed a hybrid model for linking on-prem and cloud-based SharePoint estates. This has implications well beyond AD FS.
For example, if you want to connect (let alone merge) search results, profiles, workflows, metadata models, and such across on-premise and cloud SharePoint, you typically build or rent middleware running in....Azure.
Advice for You
Hopefully, Microsoft learned all the right lessons from this outage. Let's also remember that your own infrastructure and networks can fail, too.
For you the customer, just recognize that across Yammer, SharePoint, and IaaS/PaaS services running in Azure, you're looking at three distinct environments, each with their own benefits and pitfalls. Weigh your decisions accordingly, and let us know if we can help with your assessments.