I never stand still in the cloud. Every week the Hyperscalers announce new solutions and products. Microsoft has now developed a high-availability solution for Azure regions that is designed to absorb the failure of entire data centers.
Availability Sets are used in Azure to ensure that virtual machines (VMs) of the same Availability Set are distributed across multiple fault and update domains. A fault domain is an isolated infrastructure consisting of racks, compute, network and power. VMs are thus split sequentially across multiple fault domains. If one VM fails, another VM takes over in the next fault domain. The same applies to update domains. All VMs within an update domain are booted simultaneously during patching. The strategy is to divide the VMs, which represent a cloud service as a whole, into different availability sets with fault and update domains.
With Availability Zones, Microsoft expands Azure's portfolio to protect a cloud service from a data center outage. This is achieved by isolating entire data centers. Each data center corresponds to one Availability Zone. Three of these are required at least within one region. The application architecture ensures that resources used (VMs, storage, public IPs, load balancers...) are distributed among individual availability zones. The special feature here is that only very low latencies occur in the availability zones. This does not affect the cloud service. Before Availability Zones existed, availability had to be realized across regions, but this was only possible asynchronously due to the high latencies between the regions. Availability zones increase availability to 99.99%. Provided you use at least two VMs and distribute them to different zones. Currently, this solution is only available in the Central US and France Central regions. East US 2, West Europe and Southeast Asia are in the preview.