I’ve spent a fair chunk of my career telling people the same thing: “resiliency” isn’t a checkbox you tick once and forget about. It’s availability zones, it’s backup, it’s disaster recovery, it’s knowing your RPO and RTO for every workload that matters, and it’s testing all of that regularly enough that you actually trust it when something goes wrong. The problem has always been that those pieces live in different corners of Azure - Advisor gives you recommendations, Backup Center handles backups, Site Recovery handles failover, Chaos Studio handles testing - and stitching them into one coherent picture of “how resilient am I, actually?” has traditionally meant a lot of tab-switching and a spreadsheet.
So when I started digging into the new Resiliency experience in the Azure portal, I wasn’t expecting to be this impressed. It’s genuinely one of the better pieces of platform tooling Microsoft has shipped this year, and I want to walk through why.
What’s actually new here#
What used to be the Azure Business Continuity Center has grown up into something much bigger: a unified Resiliency experience in the portal that pulls Zone Resilience and High Availability, Backup and Disaster Recovery, and Ransomware Protection into a single place. Alongside that, Azure Infrastructure Resiliency Manager has moved into public preview and is open to all Azure customers - and it’s the part I’ve enjoyed using the most.
Search “Resiliency” in the portal and you land on a dashboard with a proper menu: Overview, Resource Resiliency, Service Group Resiliency, Recommendations, Recovery Plans, Drills, and Usage Plans. That structure alone tells you the intent - this isn’t a dashboard for admiring a score, it’s a workflow that takes you from “here’s your posture” to “here’s what to fix” to “here’s how to prove it works.”
Resilience goals - finally, RPO and RTO in the platform itself#
The bit I like most conceptually is resilience goals. Instead of resiliency being a vague aspiration, you assign actual goals to a service group - your recovery point and recovery time objectives, your high availability requirements - and the platform evaluates your resources against them. It tells you plainly which resources meet the goal and which don’t, tied to real numbers rather than a generic “your resiliency score is 72%” type of output that’s hard to act on.
For anyone who’s sat in a meeting trying to get a straight answer out of a stakeholder on “what’s our actual RTO for this service,” having that recorded as a first-class object in the platform - and checked against automatically - is a small thing that solves a genuinely annoying, recurring problem.
Recommendations that don’t just tell you, they show you how#
Azure Advisor has surfaced reliability recommendations for years, but Infrastructure Resiliency Manager takes that a step further. For every resource falling short of its assigned goal, it generates a targeted recommendation with a step-by-step portal flow, the dependencies involved, and the readiness checks needed before you remediate. That’s the difference between “this VM isn’t zone redundant” and actually being walked through what changes, what it might break, and how to do it safely. Advisor tells you the gap; this tells you how to close it.
Drills - the part that made me sit up#
This is the standout feature for me. Infrastructure Resiliency Manager now supports availability zone failure drills, powered by Azure Chaos Studio under the hood. You can simulate a zone outage in a controlled way - shutting down VMs in a target availability zone, forcing failover on zone-redundant databases, or stopping AKS node pools - and watch what actually happens to your application.
Every resiliency engineer I’ve ever spoken to agrees on one thing: untested failover is a hope, not a plan. The number of environments I’ve seen where “we’re multi-zone” turns out to mean “we deployed resources across zones and never once tested what happens when a zone actually goes away” is higher than it should be. Having that drill capability built into the portal, rather than requiring a separate Chaos Studio project stood up from scratch, removes a lot of the friction that’s kept teams from testing this properly.
The resiliency agent in Azure Copilot#
There’s also a resiliency agent available through Azure Copilot now, and it’s a nice example of agentic tooling actually being useful rather than a gimmick bolted onto a product page. You can ask it things like “how secure are my backups?” and get a straight compliance summary back, with the option to have it apply the fix directly or hand you the script to run it yourself. It can review an application’s architecture against resiliency patterns - zone redundancy, failover readiness - and come back with a prioritised improvement plan, then keep evaluating your environment afterwards so gaps get flagged before they turn into incidents rather than after.
It’s not replacing the judgement calls - you still decide what gets fixed and when - but as a way of turning “we should really look at resiliency at some point” into an actual prioritised list on a Tuesday morning, it’s a good use of Copilot.
The caveats, because every post like this needs them#
- It’s public preview. Resource type coverage is still expanding - virtual machines, databases, AKS and networking are well supported, but if your estate leans on something more niche, check before you assume it’s covered.
- Drills cost real money and carry real risk. Simulating a zone failure against production is exactly as serious as it sounds. Start in a non-production environment, understand the blast radius, and get the right people in the loop before you run one against anything that matters.
- A resiliency dashboard doesn’t replace an incident response plan. This tooling makes the technical posture visible and testable, but the humans still need to know what they’re doing when the phone rings at 3am. Don’t mistake a good score for readiness.
- The Copilot agent is a force multiplier, not a decision-maker. Review what it proposes before you let it touch production, same as you would with any automation.
Final thoughts#
What impresses me most about this isn’t any single feature, it’s that resiliency finally has a home in the portal instead of being scattered across half a dozen blades that only ever get glanced at during an audit. Goals give you something concrete to measure against, recommendations turn gaps into a to-do list instead of a vague warning, drills let you actually prove your failover works instead of assuming it does, and the Copilot agent keeps an eye on all of it in between.
If you’ve got production workloads in Azure and it’s been a while since you properly reviewed how they’d cope with a zone going down, search “Resiliency” in the portal this week and have a look at what your Resource Resiliency view tells you. I’d be surprised if it doesn’t turn up at least one thing worth fixing.
