Azure Regions

Before we review Availability Zones and Availability Sets, we should review Azure regions. An Azure region is a set of closely located, well-connected data centers within a specific geographic area. These data centers are positioned close enough to offer low-latency network performance but spaced far enough apart to provide fault isolation, such as using separate power and cooling sources. This architecture enables high availability and resilience.

Availability Sets: High Availability Within a Data Center

Understanding Fault and Update Domains

Before diving into Availability Sets, it’s important to understand fault domains and update domains:

Fault Domains: Represent a group of hardware that shares a common power source and network switch. If any component in a fault domain fails, all virtual machines (VMs) within the Fault Domain will be affected.
Update Domains: Used to stagger Azure updates across the infrastructure. Microsoft updates these domains one at a time, reducing the risk of simultaneous downtime.

Availability Sets distribute your VMs across multiple fault and update domains to mitigate these risks.

How Availability Sets Work

Imagine deploying three VMs for a high-availability cluster. Without Availability Sets, it’s possible that two or more VMs end up in the same domain. If that domain fails or is updated, multiple VMs could go offline simultaneously.

Availability Sets prevent this by spreading VMs across:

Up to 3 Fault Domains
Up to 20 Update Domains

If you deploy more VMs than domains, Azure will begin to stack VMs into the same domains. For example, 11 VMs in a set with 10 Update Domains will result in one Update Domain hosting two VMs.

Note: Availability Sets must be specified at VM creation. Existing VMs cannot be added without recreating them on a new VM and attaching the original disks.

Managed Disks Consideration

If you’re using Managed Disks (which most are), Azure aligns the disk fault domains with the VM fault domains, ensuring the VM compute and disk share the same resiliency level.

Availability Zones: High Availability Across Data Centers

While Availability Sets work within a single data center, Availability Zones span multiple data centers within an Azure region.

Not all Azure regions support Availability Zones. The Azure regions that support Availability Zones contain three physically separate zones. These zones are typically within 100 kilometers of each other to support low-latency connectivity.

Benefits of Availability Zones

Each zone is an isolated failure domain.
Microsoft updates one zone at a time, similar to Update Domains in Availability Sets.
No additional data transfer costs between zones within the same region.

Deployment Models

The two types of Availability Zone deployments include:

Zonal Deployments: You choose the zone for each VM or resource (e.g., VM1 in Zone 1, VM2 in Zone 2, etc.).
Zone-Redundant Deployments: Azure automatically spreads the resource across zones. For example, a Zone-Redundant Storage (ZRS) account replicates your data across all three zones automatically.

Planning for Zones

Plan to use Availability Zones from the beginning. While migrating a resource into a zone is possible, it can be complex and varies by resource type. Azure provides a migration overview to help guide the process.

VMs can be moved into a zone using:

Redeployment
VM recreation with original disks
Azure Resource Mover

However, proactive planning is preferred to avoid disruptions.

Availability Sets vs. Availability Zones: Which to Choose?

Feature	Availability Set	Availability Zone
Scope	Within one data center	Across multiple data centers
Fault Tolerance	Limited to Fault and Update Domain	Data center-level isolation
SLA for 2+ VMs	99.95%	99.99%
Disk Alignment	Managed disks supported	Managed disks supported
Resource Migration	Recreate VM	Redeploy, recreate with OS disks, or use Resource Mover

When possible, use Availability Zones from the start, pair them with load balancing, and size your infrastructure for failover capacity.

Don’t Forget the Full High Availability Picture

Neither Availability Sets nor Zones alone guarantee high availability. Additional components include:

Load Balancers: Distribute traffic across multiple VMs.
Proper Sizing: Ensure that remaining servers can handle the load if one fails.
Service Redundancy: For example, deploying three web servers that can each handle 50% of the traffic ensures you can sustain a single-zone failure.

Simply placing VMs in different zones or sets won’t help if they can’t manage increased traffic during a failure.

Final Thoughts

Availability Sets and Availability Zones are key components in building resilient Azure solutions. They help protect against planned maintenance and unexpected outages—but they’re just one part of a well-architected cloud strategy.