We have done extensive work to improve availability of the Azure infrastructure, including innovative machine-learning to predict failing hardware early and offering premium storage to help improve reliability and performance of attached disks. As of December 12th, we are announced a new 99.9% single-instance availability SLA to better support applications that cannot easily scale beyond single VMs. We hope this enables you to move even more workloads into Azure and take advantage of the agility of the cloud without compromising on your expectations of availability.
To qualify for the single instance virtual machine SLA, all storage disks attached to the VM must be using premium storage, which offers this high level of availability and performance with up to 80,000 IOPS and 2,000 MBps of disk throughput. In addition to this new availability commitment, customers can continue to build for multi-machine high availability by having two or more VMs deployed in the same Availability Set or by utilizing VM Scale Sets which both provide machine isolation, network isolation, and power unit isolation across multiple virtual machines.
For more information about our SLA for Virtual Machines, please visit our SLA page. For additional information on premium storage and how to get started on migrating your workloads, visit the product page.