Enable zone resiliency for Azure workloads

To make your applications more resilient to availability zone-related hardware failures, network disruptions, and natural disasters, design your Azure workloads for zone resilience. When you distribute resources across multiple availability zones within an Azure region, you reduce the risk of a single zone outage affecting critical services.

Although it's a best practice to address zone resiliency during the initial planning and deployment of workloads, it's common to want to convert existing non-resilient workloads to zone resilient configurations. In general, the processing of enabling zone resiliency for existing workloads is straightforward, and Microsoft continues to simplify the process. However, any change to your workload can introduce an amount of risk. Once you understand the risks that are involved, you'll then be able to assess and prioritize which workloads and services within those workloads are most vital to your business, then apply zone resiliency to the most impactful resources first.

This article outlines key considerations for enabling zone resilience in your Azure workloads. It helps you plan, implement, and validate the changes.

Tip

If you're currently designing your workloads or plan to do a design review of your current workloads, follow the recommendations for designing for redundancy in the Azure Well-Architected Framework (WAF). The WAF recommendations guide helps you design workload redundancy across multiple levels, with a focus on critical workflows. In addition to supporting availability zone adoption, it also outlines strategies like multiregion deployments and deployment stamps.

What is zone resiliency?

Azure services can be made resilient to availability zone outages in two primary ways:

Zone-redundant services: Many Azure services support zone redundancy. These services automatically replicate data between availability zones, distribute incoming requests, and fail over to different zones during a zone failure. Each service supports these capabilities in a way that makes sense for that specific service. The mechanisms and recovery behavior vary by service, and applications can still experience transient interruptions during failover. Some services are zone redundant by default, while other services might need you to configure zone redundancy.
Zonal services: Some Azure services are zonal, which means that they can be pinned to a specific availability zone. To achieve zone resiliency by using a zonal service, deploy separate instances of the service in multiple availability zones. You might also need to manage traffic distribution, replication of data, and failover between the instances.

Some services can be deployed in either a zone-redundant or zonal configuration. For most cases, it's best to deploy zone-redundant services when you can.

For more information, see Types of availability zone support.

Zone resilience enablement procedure

Use the following steps to systematically review your Azure workloads, prioritize them for zone resiliency, and enable zone resiliency for each component.

Prerequisites

Before you begin, perform the following actions:

Identify each workload. A workload refers to a collection of application resources, data, and supporting infrastructure that function together to achieve defined business outcomes. For more information about workloads and how to define them, see Well-Architected Framework workloads.
Prioritize each workload's user and system flows. Understand the critical paths and dependencies of your workloads to determine which components to make zone-resilient first. For more information about how to use critical flow analysis to prioritize workflows, see Architecture strategies for identifying and rating flows.

Assign a criticality rating to each workload and flow. This rating helps you understand the impact of a potential outage on your business and guides your decisions about which workloads to prioritize for zone resiliency. Also consider the amount of acceptable downtime while you reconfigure the workloads.

You can use a taxonomy to classify your workloads based on their criticality. This approach helps you focus your efforts on the most important services.

Consider the following example taxonomy to classify your workloads.

Workload type	Description	Effect of disruption
Mission-critical	Critical flows and workloads that must be highly reliable, always available, resilient to failures, and recoverable within defined recovery objectives if any downtime occurs.	Any disruption to essential functions immediately risks catastrophic business damage or introduces risks to human life.
Business-critical	Essential flows and workloads that operate important business functions.	Disruption risks some financial loss or brand damage.
Business‑operational	Contributes to efficiency of business operations, but out of direct line-of-service to customers.	Can tolerate some level of disruption.
Administrative	Internal production flows and workloads not aligned to business operations.	Can tolerate disruption.

For more information about how to classify your workloads according to criticality rating, see Assign a criticality rating to each flow.

Verify that the regions where your Azure resources reside support availability zones. Consult the Azure regions list. If a region doesn't support availability zones, consider relocating your resources to a region that does. For more information, see Move Azure resources across resource groups, subscriptions, or regions.

Step 1: Prioritize Azure services for zone resilience

After you determine which workload flows are most critical to your business, you can focus on the Azure services that those flows depend on. Some Azure services are more critical to your applications than others. Prioritize these services to help ensure that your applications remain available and resilient if a zone failure occurs.

Use the following guidance to prioritize Azure service groups based on their criticality to your workloads. Consider your specific application architecture and business requirements when you determine the priority of services for zone resiliency.

Start with networking services. Workloads tend to share networking services, so an increase in the resiliency of these services can improve the resiliency of multiple workloads at once.

Many core networking services are zone redundant automatically, but you should focus on components like Azure ExpressRoute gateways, Azure VPN Gateway, Azure Application Gateway, Azure Load Balancer, and Azure Firewall.
Improve operational data storage resiliency. Operational data stores contain valuable data that multiple workloads often use, so improving the availability of those data stores can help many workloads.

For operational data storage resiliency, focus on services like Azure SQL Database, Azure SQL Managed Instance, Azure Storage, Azure Data Lake Storage, Azure Cosmos DB, Azure Database for PostgreSQL, Azure Database for MySQL, and Azure Managed Redis.
Prioritize compute services. Stateless application instances can be replicated and distributed across zones without replicating application state between the instances. However, compute services can also host stateful applications. Account for their data, storage, and replication requirements when you plan zone resilience.

Compute services include Azure Virtual Machines, Azure Virtual Machine Scale Sets, Azure Kubernetes Service (AKS), Azure App Service, App Service Environment, Azure Functions, Azure Service Fabric, and Azure Container Apps.
Review remaining business-critical resources that your critical flows use. These resources might not be as critical as the resources listed previously, but they still play a role in your application's functionality, and you should consider them for zone resiliency.
Review the rest of your business-operational resources. Make informed decisions about whether to make them zone resilient. This review includes services that might not directly relate to your critical workloads but still contribute to overall application performance and reliability.

Step 2: Assess zone configuration approaches

After you prioritize your workloads and Azure services, identify the approach required to enable availability zone support for each service. Understand what you need to do to configure zone resiliency.

Each Azure reliability service guide provides a section that describes how to enable zone resiliency for that service. This section helps you understand the effort required to make each service zone resilient so that you can plan your strategy accordingly. For more information about a specific service, see Azure reliability service guides.

Use the zone configuration table to identify approaches for common Azure services.

Important

If your workload includes components deployed in a zonal (or single-zone) configuration, plan to make these components resilient to zone outages. A common approach is to deploy separate instances into another availability zone and switch between them if necessary.

Step 3: Test for latency

When you make workloads zone resilient, consider latency between availability zones. Occasionally, some legacy systems can't tolerate the small amount of extra latency that cross-zone traffic introduces, especially when you enable synchronous replication across zones within the data tier. If you suspect that cross-zone latency might affect your workload, run tests before and after you enable zone resiliency. For more information about how cross-zone latency might affect your application and approaches to mitigate cross-zone latency problems, see Zonal resources and zone resiliency.

Step 4: Implement the changes

For each component, follow the configuration or migration instructions in its reliability guide. Before you change production resources:

Define acceptance criteria for your critical flows, including acceptable downtime, latency, throughput, and data loss.
Check the service's region, tier, capacity, and quota requirements. Include dependent resources and any temporary resources that the migration requires.
Confirm that you have recoverable backups where applicable. Use the reliability guide's Backup and restore section to understand how to back up each resource. Document how to roll back or recover if the change fails, including how to handle data written after a migration cutover.
Test the change and recovery procedures in a representative nonproduction environment.

Apply changes in stages where the service supports it. Monitor application health and critical flows during each stage, and use your rollback or recovery procedure if the results don't meet your acceptance criteria.

Step 5: Validate zone resilience

After you apply the changes, verify that the workload meets the acceptance criteria you defined. Validate the new configuration, such as by checking the service is now configured for zone redundancy.

For some services, you manage the number of instances. Check the service's instances are distributed among zones. Assess whether the capacity in the surviving zones can meet your requirements if one zone becomes unavailable. Don't assume that you can allocate replacement capacity during an outage. For more information, see Manage capacity with over-provisioning.

For services that support failure testing, follow its reliability guide to test application behavior during failure and recovery in a controlled environment. Test the behavior that you're responsible for, rather than trying to validate Microsoft's internal failover mechanisms.

Zone configuration approaches for Azure services

Each Azure service supports a specific type of availability zone support, which is based on the service's intended use and internal architecture. If you have a resource that isn't configured to use availability zones (also called a nonzonal resource), you might want to reconfigure it with availability zone support. The reliability guide for that service provides guidance or links to availability zone configuration instructions.

This section provides an overview of the different types of zone configuration approaches and which approach each service supports. This section excludes nonregional services because Azure doesn't deploy them to a single region. For more information about nonregional services, see Azure nonregional services and the service's reliability guide.

Important

When you enable zone redundancy on a resource, that resource becomes automatically resilient to zone failures. When you use a zonal configuration to pin the resource to a specific availability zone, the resource isn't automatically zone redundant. You must make it resilient to a zone failure. For zonal services, this article reflects the complexity and cost of pinning to a single zone, not to make it zone-resilient. For more information about extra steps to achieve zone resiliency, see the service's reliability guide as well as Zonal resources and zone resiliency.

The zone configuration table lists the supported zone configuration approach for many Azure services and contains a link to each reliability guide for that service. The reliability guide provides information about how to configure nonzonal service resources to enable availability zone support.

The following table describes common zone configuration approaches, including the level of effort and downtime required to enable availability zones. Some services have different approaches listed because of the way they work.

Approach	Description	Typical level of effort	Might require downtime
Always zone resilient	The service is zone resilient by default in regions that support availability zones. No action is required.	None	No
Enablement	Minimal configuration changes required, such as enabling zone redundancy in settings. The process has no or minimal effect on the resource's availability, but consider effects on cost or performance.	Low	None or minimal
Modification	Likely requires some configuration changes, such as redeploying dependent resources or modifying network settings.	Medium	Yes
Redeployment	Significant changes required, such as redeploying entire resources, applications, or services, or migrating data to new services.	High	Yes

Understand the cost of enabling availability zone support for a service. Some services don't charge separately for the feature, but you still pay for the resources and capacity that you deploy. Other services require a specific tier, a minimum number of instances or capacity units, or both. For storage services, zone-redundant storage (ZRS) can have higher rates than locally redundant storage (LRS). The table in the next section identifies these cost factors.

Note

The information in this article summarizes the typical approach to enable availability zone support and outlines the typical cost impact. But some factors might affect how it works for your specific solution. For example, some services are listed as always zone resilient, but this designation only applies in specific regions or for specific tiers of the service. Use these tables as a starting point, but review the linked reliability guide to understand the specific details.

Azure services by zone configuration approach

The following table summarizes the availability zone support for many Azure services and provides an approach, including cost impact, to enable availability zone support for each service.

Service	Typical zone configuration approach	Typical cost impact
Azure AI Search	Enablement	Minimum tier and replica count required
Azure API Management: Premium (classic) tier	Automatic configuration: Always zone resilient Manual zone reconfiguration: Modification	Additional units billed at Premium tier rates
Azure API Management: Standard v2 and Premium v2 tiers	Redeployment	Additional units billed at the selected tier's rates
Azure App Configuration	Always zone resilient	N/A
Azure App Service	Enablement	Minimum tier and instance count required
Azure App Service - App Service Environment	Enablement	Minimum instance count required
Azure Application Gateway	Always zone resilient	N/A
Azure Application Gateway for Containers	Always zone resilient	N/A
Azure Automation	Always zone resilient	N/A
Azure Backup	Redeployment	Moderate cost increase
Azure Bastion	Redeployment	No cost impact
Azure Batch	Redeployment	No cost impact for same number of virtual machines (VMs)
Azure Blob Storage	Enablement	Moderate cost increase
Azure Chaos Studio	Always zone resilient	N/A
Azure Container Apps	Redeployment	Minimum replica count required
Azure Container Instances: Container groups	Redeployment	No cost impact
Azure Container Instances: NGroups (preview)	Redeployment	No cost impact for same number of container groups
Azure Container Registry	Always zone resilient	N/A
Azure Cosmos DB	Modification	None if using autoscale or multiregion writes
Azure Data Explorer	Modification	Moderate cost increase
Azure Data Factory	Always zone resilient	N/A
Azure Data Lake Storage	Enablement	Moderate cost increase
Azure Database for MySQL	Redeployment	Requires primary and high availability (HA) replica
Azure Database for PostgreSQL	Enablement	Requires primary and HA replica
Azure Databricks	Enablement	No cost impact for same number of VMs; moderate cost increase for storage
Azure Disk Storage	Enablement	Moderate cost increase
Azure DDoS Protection	Always zone-resilient	N/A
Azure Device Registry	Always zone resilient	N/A
Azure DNS Private Resolver	Always zone resilient	N/A
Azure DocumentDB	Enablement	Requires primary and HA standby shard
Azure Elastic SAN	Redeployment	Moderate cost increase
Azure Event Grid	Always zone resilient	N/A
Azure Event Hubs: Dedicated tier	Always zone resilient	Minimum capacity units (CUs) required
Azure Event Hubs: all other tiers	Always zone resilient	N/A
Azure ExpressRoute gateway	New gateways: Zone resilient by default Existing nonzonal gateways: Modification	Minimum SKU required
Azure Files	Enablement	ZRS: Moderate cost increase Zonal LRS: No cost impact
Azure Firewall	New firewalls: Zone resilient by default Existing nonzonal firewalls: Modification (automatic migration in progress)	No cost impact
Azure Functions: Dedicated plan	Enablement	Minimum instance count required
Azure Functions: Flex Consumption plan	Enablement	Minimum instance count required
Azure Functions: Premium plan	Redeployment	Minimum instance count required
Azure HDInsight	Redeployment	No cost impact for same number of nodes
Azure IoT Hub	Always zone resilient	N/A
Azure Key Vault	Always zone resilient	N/A
Azure Kubernetes Service (AKS)	Redeployment	No cost impact
Azure Load Balancer	Modification	No cost impact
Azure Logic Apps - Consumption tier	Always zone resilient	N/A
Azure Logic Apps - Standard tier	Redeployment	Minimum tier and instance count required
Azure Managed Grafana	Redeployment	Moderate cost increase
Azure Managed Redis	Enablement	Requires primary and HA instance
Azure Monitor health models	Always zone resilient	N/A
Azure Monitor Logs	Always zone resilient
Azure NAT Gateway	Redeployment	No cost impact
Azure NetApp Files	Redeployment	Depends on replication configuration
Azure Notification Hubs	Redeployment	Moderate cost increase
Azure Private Link service	Always zone resilient	N/A
Azure Queue Storage	Enablement	Moderate cost increase
Azure Service Bus	Always zone-resilient	N/A
Azure Service Fabric	Redeployment	No cost impact for same number of VMs
Azure SignalR Service	Enablement	Minimum tier required
Azure Site Recovery	Redeployment	No cost impact for Site Recovery, moderate cost increase for replica storage
Azure SQL Database: Business Critical tier	Enablement	No cost impact
Azure SQL Database: General Purpose tier	Enablement	Moderate cost increase
Azure SQL Database: Hyperscale tier	Redeployment	Minimum replica count required
Azure SQL Database: Premium tier	Enablement	No cost impact
Azure SQL Managed Instance	Enablement	Moderate cost increase
Azure Storage Actions	Always zone resilient (control plane) Enablement (target storage accounts)	N/A (control plane) Depends on target storage account redundancy
Azure Stream Analytics	Always zone-resilient	N/A
Azure Storage Discovery	Always zone-resilient	N/A
Azure Table Storage	Enablement	Moderate cost increase
Azure Virtual Machine Scale Sets	Redeployment	No cost impact for same number of VMs
Azure Virtual Machines	Modification	No cost impact for same number of VMs
Azure Virtual Network	Always zone resilient	N/A
Azure Virtual Network Manager	Always zone-resilient	N/A
Azure VMware Solution	Redeployment	No cost impact for same number of nodes
Azure VPN Gateway	Modification	Minimum SKU required
Azure Web PubSub	Enablement	Minimum tier required
Public IP address	Always zone resilient	N/A