availability zones

What are availability zones (AZs)?

Availability zones (AZs) are isolated or separated data centers located within specific regions in which public cloud services originate and operate. Cloud computing businesses typically have multiple worldwide availability zones. This helps ensure cloud customers have a stable connection to a cloud service in the geographic AZ that's closest to them.

Cloud service providers (CSPs) host their resources and data centers in multiple locations worldwide. The locations that are isolated from each other but close enough to have low-latency connections with each other are known as availability zones. AZs represent parts of regions, and each AZ includes one or more data center.

Isolation and distance between AZs reduces the probability that more than one AZ will be affected by adverse weather conditions, power outages or some other disaster in that geographic location. Thus, even if one AZ goes down due to a disaster, the remaining AZs will continue to support regional services, provide computing and storage capacity, and ensure high availability for the cloud's users. At the same time, the distance between the AZs is short enough to ensure that the connection between the AZs offers a very low round-trip latency (typically of a few milliseconds) for any necessary data transmissions.

The larger CSPs like Amazon Web Services (AWS) and Microsoft Azure typically offer more cloud computing resources and therefore have multiple worldwide AZs. This helps ensure that cloud customers can access the resources they need in the geographic location that's closest to them via stable, low-latency connections.

Data centers in availability zones

An AZ consists of multiple data centers, which are all equipped with independent power, cooling and networking infrastructure, and are separate from the data centers in other AZs. The data centers within each AZ are also located apart from each other, again to minimize the potential adverse impact of an outage and to ensure continued availability and capacity.

Data centers in an AZ connect to each other over a redundant, high-speed, low-latency private network link, and all AZs in a region connect through the same sort of network links. The most reliable CSPs select data center locations very carefully and only after conducting a thorough vulnerability assessment. This assessment allows them to identify data center-specific risks and also understand if any risks will be shared between AZs.

What are the benefits of availability zones?

Organizations select AZs for a variety of reasons, including compliance and proximity to customers. Cloud administrators can also choose to replicate services across multiple AZs to decrease latency or protect resources. Admins can move resources to another AZ in the event of an outage. Specific cloud services may also be limited to particular regions or AZs.

When an organization launches a cloud instance, it selects a region and then an AZ. Normally, there are multiple AZs that customers can choose from in a region, or they can let the cloud service select their zone. Distributing instances across multiple AZs provides redundancy and failover, so that even if one instance fails, an instance in another AZ can handle requests (as long as the application is designed to support the failover). If one data center in an AZ encounters a problem, the cloud provider can still offer its services through the remaining zones.

What is a multi-AZ deployment?

AWS provides a managed offering called multi-AZ deployment to ensure high availability and failover support for database (DB) instances. With this offering, AWS users get one or two standby DB instances in a different AZ. If there is an AZ failure and the master/primary DB in that AZ goes down, the multi-AZ deployment ensures that the standby instance provides failover support. The primary DB instance is synchronously replicated across availability zones to the standby replica (or replicas) to provide data redundancy, minimize latency spikes, ensure high availability and protect the DB against DB instance failure during planned system backup or maintenance.

A multi-AZ deployment with one standby instance (known as a multi-AZ DB instance deployment) provides failover support, but doesn't serve read traffic. Also, there is only one row for the DB instance.

On the other hand, a multi-AZ deployment with two standby instances (known as a multi-AZ DB cluster deployment) can do both. In this type of multi-AZ deployment, there is a cluster-level row with three DB instance rows under it.

Regions vs. availability zones vs. local zones

Availability zones are multiple, isolated locations within a region. Each AZ is identified by a unique code. For example, the code for AWS' AZ named US West (N. California), launched in 2009, is us-west-1; the code for the AZ named Asia Pacific (Mumbai) that was launched in 2016 is ap-south-1.

Some CSPs allow customers to use multiple AZs together. By doing so, they can maintain separate copies of their applications and data within different data centers that are physically separated from each other in a large metropolitan area.

The AZ resources available to customers can be either pinned to a specific AZ (zonal resources) or spread across multiple AZ (zone-redundant resources). Customers who choose zonal resources are responsible for failover to another AZ if an outage occurs in one AZ. The CSP manages failovers for zone-redundant resources.

A region is a geographical location that includes multiple AZs. Cloud providers typically have two or more AZs within each region. Every region is isolated and independent from other regions. While one region may have multiple AZs, no AZs are shared among different regions. Regions are spread out all over the world, so cloud providers can reach customers on multiple continents.

Some CSPs also offer local zones in some geographic regions. AWS for example, offers six local zones in the US West (Oregon) region, two local zones each in the Mumbai, Sydney, Singapore and Europe (Stockholm) regions, and so on. A local zone is an extension of a region that allows cloud customers to place compute and storage resources in a region closer to their users and to take advantage of low-latency communications. With AWS, customers can extend a virtual private cloud to a local zone by creating a subnet in that local zone.

Availability zones in AWS, Azure and Google Cloud

The big three CSPs -- AWS, Microsoft Azure and Google Cloud -- use both regions and availability zones. That said, Google uses the term "zones" rather than "availability zones" to refer to the locations of its cloud resources.

The availability zones each CSP offers are also independent, so a failure or outage within one AZ won't affect other zones; the other zones will pick up the slack from the region that went down. This ensures that an organization's disaster recovery plan isn't compromised.

As of 2024, the regions and AZs in AWS, Azure and Google Cloud are as follows:

• Azure has more than 40 regions, with many regions supporting one or more AZs. Several regions and AZs are also in development. For each Azure region, there are multiple data centers that can run independently. Azure regions are spread across Africa, Asia Pacific, Canada, Europe, Middle East, South America and the U.S.
• Google Cloud spans 40 regions and 121 zones spread across over 200 countries and territories. Eight additional regions will be added to Google Cloud beginning in 2024.

• AWS has 33 regions and 105 availability zones, plus 38 local zones, together serving 245 countries and territories. An additional 12 AZs and four regions are under development.

Choosing an availability zone

When choosing an AZ from a cloud provider, users should consider the following:

• Latency and proximity. The closest zone will offer the lowest latency.
• Compliance. Depending on the location, cloud providers may need to meet different regulatory compliance laws and regulations, such as with the GDPR, so it's important to choose an AZ that meets the compliance requirements that apply to the user organization.
• Service-level agreement for availability. The level of availability may change depending on the cloud provider or region. Normally, the AZs of AWS, Azure and Google Cloud provide high availability.
• Cost. Different regions may cost more than others, and the cost may change by cloud provider.
• Redundancy and failover. The choice of CSP and AZ would also depend on whether the organization needs data or apps stored in one location or across multiple locations.

Learn how AWS Regions and Availability Zones differ, and how they affect cloud costs and configurations. Compare AWS, Azure and Google Cloud identity and access management services and explore best practices to achieve high availability in cloud computing.