Zonal Resiliency in Azure: Application-Centric Goals, Recovery Plans, and Drills

Hello, everyone!

If you’ve ever looked at a multi-tier app in Azure and wondered, “Will this really hold up during a zone outage?”, you’re not alone. During session MAIS23 at the Microsoft Azure Infra Summit 2026, Bhavya, Aditya, and Chaya from the Azure Resiliency team shared insights about the newly rebranded Resiliency in Azure experience (previously known as the Azure Business Continuity Center). They emphasised the importance of viewing resiliency not just as a checklist item for individual resources, but as a comprehensive outcome for the entire application.

Most of us have been there. An app is hosted “in the cloud”, spanning IaaS VMs, PaaS databases, an app service plan, and a shared Azure Firewall managed by another team. But when a zonal glitch occurs, suddenly it’s hard to determine: was this app meant to be zone resilient or not?

The session began with the story of Zava, a rapidly growing insurance firm running a claims app with an impressive 99.9% availability. Unfortunately, they lost over $40,000 in revenue in just one week due to zonal outages. That’s a significant loss, highlighting a common issue many organisations face.

Here’s why this is vital for IT professionals:

• You get a clear overview of the zonal resiliency status across IaaS, PaaS, and shared services.
• Resiliency targets are set at the application level, rather than being hidden within individual resource settings.
• Receive customised Azure Advisor recommendations, along with an Azure Copilot guided flow that generates remediation scripts.
• Run zone-down drills using Azure Chaos Studio without having to mix and match multiple tools.
• Recovery plans streamline failover in a specific order, along with on-demand readiness checks ahead of any possible outages.

In short, you’ll spend less time guessing, save on tedious spreadsheet tracking, and have much more confidence that your app will perform as promised to the business.

The team has also rebranded the Azure Business Continuity Center to Resiliency in Azure. This new unified solution addresses infrastructure, data, and cyber resiliency all in one place. Our current focus is on zonal resiliency, with plans for regional disaster recovery and clear RPO/RTO goals in the pipeline.

The central idea here is the service group. This is a logical unit that spans across subscriptions and resource groups. You incorporate VMs, databases, app service plans, Redis caches, and all other Azure resources needed for your application. From there, resiliency operations can be managed collectively, rather than resource by resource.

You’ll mainly work within these two views:

• Resource resiliency: A summary view of zonal configurations across approximately 20 resource types currently supported.
• Service group resiliency: A similar summary, but focused on the application level to help you prioritise which apps need attention first.

The speakers were candid about the current scope. Right now, objectives are straightforward; it’s about determining if “this service group should be checked for zonal resilience.” Future expansions will include additional pillars such as regional disaster recovery and associated goals for RPO and RTO. I appreciated their honesty in not overselling the solution.

Once you create a service group, there are three key components in the workflow. Each is designed to address challenges many of you might be facing:

1. Goals and recommendations: Assign a zonal resiliency goal to the service group, and Azure Advisor will provide tailored recommendations based on its resources. Here are a couple of noteworthy points:

• The view clearly displays the cost implications before you implement any changes. Some Azure services incur no costs for zone redundancy, while others do. You can see this information directly rather than navigating to a separate calculator.
• There’s an Azure Copilot guided remediation workflow that walks you step-by-step through the recommendations to produce a script, which accommodates various resource-type scenarios (like SKU changes or redeployments) meant for your automation pipeline.

You can also exclude a resource if it’s not critical (“zonal redundancy not needed”) or manually attest to a resource when your custom solution already provides the necessary resiliency that the platform doesn’t automatically detect. This option is crucial, as real-world environments often feature a few unique cases.

2. Application-centric recovery plans: Instead of failing over individual resources one by one, recovery plans manage the entire application. They can auto-detect existing services (like Azure Site Recovery for VMs), allow you to group and sequence resources for failover, and can exclude resources already configured for high availability (there’s no need to fail over something that’s still operational). You can perform an on-demand readiness check whenever the application structure changes to catch any configuration drift before an outage arises.
3. Zone-down drills powered by Azure Chaos Studio: A zone-down drill template identifies the resources within the service group, automatically sets the right faults for each resource type (like a Redis cache issue or a VM scale set shutdown), and includes identity checks, monitoring, and your established recovery plan. When executing, you choose the region and target zone, undergo pre-validation, inject the fault, initiate failover, then return to protection—all logged as a single operation. Metrics per resource help to illustrate the downtime for each component involved. If the default fault isn’t suitable, you can modify it with a custom runbook.

This last point deserves emphasis: A drill isn’t just for fault injection. It combines fault injection, failover, reprotection, and failback, all tracked and verifiable in one cohesive space.

Returning to Zava, they needed answers to three questions: what is our current zonal resiliency status across these Azure services, what should we prioritise to maintain our 99.9% target, and how do we verify that we’ll actually function during an outage? Resiliency in Azure provides answers to all these inquiries without the need for the platform team to write extensive PowerShell scripts.

Here are some use cases worth considering:

• Regulated workloads: Industries like insurance, healthcare, and finance that must demonstrate compliance with drills. Avoiding the hassle of manual attestation was clearly designed for auditors.
• Mixed estate applications: These are where a central team manages shared services (like firewalls and identities) while app teams handle the rest. Service groups can reflect this organisational structure.
• Custom resiliency solutions: Apps that have their own solutions which the platform can’t detect may use manual attestation to keep the dashboard accurate without needing a complete overhaul.
• Game-day rehearsals: Thanks to the pre-built zone-down template, you can conduct a meaningful drill in just one afternoon instead of needing to design a completely custom Chaos Studio experiment from scratch.

The reality is that achieving zone redundancy isn’t free for every service and not every resource type is currently supported (around 20 at present). Plan wisely, exclude non-critical assets, and attest to those covered by other solutions.

If I were to tackle this on a Monday morning, here’s the approach I’d take:

1. Log into the Azure portal and search for Resiliency. You’ll find the revamped Resiliency in Azure page that took over from the old Business Continuity Center.
2. Create a service group. Add resources directly, or include entire resource groups if they already align with application boundaries in your setup.
3. Assign the zonal resiliency goal to this service group.
4. Review the summary tiles. Exclude or attest the necessary resources.
5. Examine the Advisor recommendations. Use the Copilot guided flow to generate a remediation script and run it through your automation system.
6. Forge an application-centric recovery plan. Group and order the resources, then execute an on-demand readiness check.
7. Create a zone-down drill from the template, ensuring identity, monitoring, and fault validations are in place, and then implement the drill in a non-production zone first.

View the complete Microsoft Azure Infra Summit 2026 session playlist here: https://www.youtube.com/playlist?list=PLjt5SKzX1iI8con7FJDB56G6hHqxGm7ki

Cheers!

Pierre Roman