There's been a lot of hubbub recently over Microsoft Azure outages. And, for good reason, considering that over the past couple weeks, Azure customers experienced 6 out of 9 days of outages. The outages affected different services at varying times for multiple regions, but Monday's (August 18th) outage was the most severe affecting Cloud Services, Virtual Machines Websites, Automation, Service Bus, Backup, Site Recovery, HDInsight, Mobile Services and possible other Azure Services in multiple regions for about 5 hours.
Obviously the outage on Monday dramatically exceeds Microsoft's Service Level Agreements (SLAs) and it's yet to be seen if customers will demand service credits. They should. But, despite the highly publicized outages, it's interesting to hear that Azure services in Q2 are actually much improved over Q1 this year.
CloudEndure, a Cloud continuity company, keeps track of Azure outages and service availability.
According to recent reporting, Azure experienced 201 service issues in Q2 compared to 259 in Q1, resulting in a 22 percent improvement. However, the issues experienced in Q2 were more severe, with an 800% increase in service interruptions. So, when you compare Q1 and Q2 side-by-side, there were fewer issues, but the issues that did transpire were more critical (full outages) and lasted longer.
And, that brings up a good track of thought. What can customers actually live with? Can they live with services "burps" here and there as long as outages are short?
The top 10 (in order) out of 23 tracked products experiencing problems (interruptions and degradations) were:
- SQL Database (Degradation = 37; Interruption = 8)
- Compute (Service Management) (Degradation = 27)
- Compute (Degradation = 13; Interruption = 1)
- Storage (Degradation = 11; Interruption = 1)
- HDInsight (Degradation = 12)
- Virtual Machines (Degradation = 4)
- Service Bus (Interruption = 7)
- SQL Reporting (Interruption = 8)
- Visual Studio Online (Degradation = 4; Interruption = 1)
- Cloud Service (Degradation = 2)