GLOBAL_Cayosoft Logo_Main_Full Color_Small

The New Reality of Microsoft Identity Recovery: Entra ID, Hybrid Identity and Operational Resilience

Transcript

Drew Fuhrman 0:06
All right. Hey, welcome everyone, and thanks for joining us for The New Reality of Microsoft Identity Recovery. Today's discussion is about a shift a lot of our organizations are living through right now. Microsoft 365 has become central to how we all work and get things done, and identity has really become the foundation underneath all of that. When identity is healthy, everything feels normal. But when it's misconfigured, unavailable, or difficult to recover, the impact can show up across email, collaboration apps, and the devices people are bringing into the organization, and administration falls apart very quickly.

We have two great speakers here today, Craig and Merill, and they're going to help us unpack what that really means in practical terms: where the risks are, what recovery capabilities exist today, where some of the gaps might be hidden, and what teams can do to become even more resilient. So with that, let's pop right in.

Let's start with a big-picture question. Microsoft identity has been around for a long time, and that means a lot of teams think of it as mature, stable infrastructure. But the dependency profile has changed a lot. More services, more users, more applications, and more admins mean a lot more security control now, and that flows through identity more and more. That means recovery plans that were originally designed to handle this aren't really applicable anymore, are they?
Read More

So, Merill, big question for you. When you look at the evolution of identity from the early cloud to Azure AD and Azure AD Connect, Conditional Access, and now Entra ID, what's changed the most in terms of business dependency?


Merill Fernando 2:17
Hey everyone, I'm super excited to be here. I've never been up in this time zone, so it's always fun, because it's like 3 a.m. for me here. I'm usually asleep at this time. So good to be here with you and Craig.

Yeah, so it's actually been a journey even with Microsoft Entra. If you look at the history of Entra, I remember a few years ago, not many years even, this was in the last five years, one customer, a mining company in Australia, called in a panic. They had written and run a script that accidentally started deleting all their service principals. They called Microsoft asking, what do we do now? How do we restore it? And unfortunately, there was nothing Microsoft could do to help them. This was before things like the recycle bin existed.

In the beginning, everyone was like, oh, we're in the cloud. We don't need backups anymore. We don't need to worry about any of this stuff. Microsoft will take care of it. But the reality that a lot of people figured out a little too late is that you still had responsibility for your estate. And we've seen this journey at Microsoft. Microsoft has been adding things like recoverability and the recycle bin, so now things go into the recycle bin for 30 days and you can recover them. We're seeing all of that evolve on the Entra side too, with new features like tenant governance that came out this year. I'm sure we'll dig into all of that later.


Drew Fuhrman 4:12
Yeah. And Craig, a follow-up question from an operations and recovery standpoint: where do you see organizations still treating identity as less critical than it's really become at this stage?


Craig Birch 4:14
Let's unpack a couple of things before we get to that question, because I think Merill said a couple of important things. Entra has kind of been on its own path, right? It started in, I think, 2008, and we're here now in 2026 and we're still talking about Entra ID recovery. That makes me scratch my head right away. I believe the first instance of Entra was around BPOS back in the day, and Merill, you could probably keep me honest there. That was really to provide some core services into M365.

But if we look at the evolution of Entra ID, it's way beyond that. Entra ID is a full IdP, but it's beyond just an IdP. It has conditional access policies. It's really the identity fabric in an organization.

So, to answer your question about some of the challenges and why we're seeing them, let's step back and look at Active Directory and its journey. Active Directory is what, 23-plus years old? Entra, I think, is 17 now, so they're very similar. Organizations would say, OK, I'm good because I have multiple domain controllers. If I lose a domain controller, I'm good. And they thought that was recovery until more recently. It was good enough. We have redundancy. We have multiple domain controllers. If one goes down, they're cattle, not pets. We just stand up a new one and we're up and running.

Now let's take that same principle and apply it to SaaS. It's even more likely that the organization says, well, Entra ID never goes down, the services are up, so we're good. And Microsoft owns it. It's not me.

When I owned Active Directory, I had a lot of skin in the game, because they were my servers. They were in my data center. There was real ownership. I'm not saying there's no ownership in SaaS, so don't take this the wrong way. But as I move my identity to the cloud, I feel less ownership. If something goes wrong, it's Microsoft's responsibility. And Merill, you said it: they were expecting that Microsoft had them covered, and that's not the case. So that's a couple of things I feel. Am I off there?


Merill Fernando 7:31
Well, this came up quite a lot when I was at Microsoft. So much, in fact, that if you search for Entra resilience, you'll end up on a Microsoft Learn page with a diagram that says, this is your responsibility and this is Microsoft's responsibility. Because you still own all the objects in your tenant. You have permissions and you have write access. So if you give too broad an access, or if an attacker gets in and wipes things and makes changes, you are still responsible for the side of the stack that you manage.

Just to make it clear for everyone, that's a published document on Microsoft Learn, and we'll share the link to it. It makes it a lot clearer where the responsibility lies. Microsoft of course does internal backups and all of that, but you are responsible for the things that you can change.


Craig Birch 8:34
See, that is the big one. So what can we change? If we go back to the evolution, it was primarily users, groups, some settings, mailbox permissions, SharePoint. But that's no longer the case. We have Entra ID applications, we have agents now inside of Entra ID, we have conditional access policies. The fabric has also moved into device management, so we now have Intune configuration. There are a lot of moving pieces.

I'm going to go one step back, and then we can move on to the next topic. If we go back to Active Directory and how organizations traditionally tried to solve this, Microsoft put out guidance: here's how you recover things. It was based on, OK, I'm going to run through this process, and now I'm back up and running. But I controlled all of it. I had some control over that.

Here, we're going to Microsoft and saying, hey, we're expecting all of these things to be part of that. If I go back to Active Directory, when I tried to back it up, I thought about it the way I'd traditionally back up a SQL server or a web server. It wasn't the same. And it's not the same for Entra ID either. It's not just understanding my users and my groups. There's a lot of tenant configuration, application settings, et cetera, that are often overlooked as part of that. And again, most organizations think Microsoft takes care of it for them.


Drew Fuhrman 10:33
Well, here's a question for you both. When we're talking about the things the organization can't control, if I'm an executive, what's the business risk if I'm not planning for recovery? Why should I even care? Somebody else should take care of that, right? Maybe it's Microsoft, maybe it's our own people. Why should we plan for that?


Craig Birch
10:58
Because identity controls everything.


Merill Fernando 11:02
Yes. The example I gave earlier is a good one. Someone deleted the apps, and now none of the users can sign in to the apps. They can't do their work, and your business is losing money. In a mining org, they work to schedules, they have deadlines, and people have to be at certain places at certain times and do things. It's the same with any business. They start losing money if people can't get to their work, because Entra and AD are the gateway to go in and get work done.

If you can't do that, you might have all the resilience for all the apps inside, and they might all be up and running. But if your users can't get to these apps, all the investment you put into making sure you had high availability for those applications, all the money you spent on them, is not much use. So that's a very critical part of this.

And people usually only realize this when something like this happens. A lot of orgs don't even realize there's an identity provider, Entra ID, in the cloud. They don't think much about it, because it's just there and it just works. It's just a portal they go to configure stuff. It's only when they hit that first outage related to it that they go, oh, we have this thing that's the center of our world.


Drew Fuhrman 12:51
Yep, and that's a really good point, Merill, because it's a good transition into the next topic: Microsoft 365 availability starts with identity. Identity is not just another service in the stack. It's the access path for almost everything people depend on every day. If authentication, authorization rules, device rules, or policy enforcement break, the service may still be running, but users and admins may not be able to reach it.

So Craig, when identity is unavailable or misconfigured, what are some of the first business functions that typically feel the impact? I know we talked about it a little earlier, but can you walk through what's breaking and how we start to see it?


Craig Birch 13:48
Absolutely. Obviously, if it's a major identity outage, the first thing they're probably going to notice is normal day-to-day operations. Their email communications: I can't access Exchange Online. I can't get to my Teams application. I can't access SharePoint, or now maybe even Copilot, depending on how the business functions.

But it's beyond that. Those are probably the first things they'll experience, but if they're really suffering a true identity outage in Entra ID, it could extend into business applications beyond the productivity suite. They could be using Entra ID for authentication as SSO into other organizations. So it really depends on what they're using Entra ID for, because Entra ID gives me a lot of capabilities that standard Active Directory did not. They're not the same.

The other piece is admins not being able to perform security tasks. They might not be able to log in to the portal themselves to validate that things are working if identity is down. So we're in a conundrum: we need identity to get to the portal, and if identity is broken, we can't get to the portal. I hate to say it, but there are a lot of moving parts inside Entra ID. I'm not saying that isn't true of other identity providers, but I think the ecosystem is very complex.


Merill Fernando 15:41
There's actually a fun story. Not fun at the time, but this was just before 2020, around 2018 or 2019. Entra used to have a few outages, and some of them were global. Teams was the biggest app, and I'd say it's still the biggest app for Microsoft, because everyone's work lives in Teams. We're on Teams right now. Without Teams, there's no M365. And Satya said no other app should ever cause Teams to become unavailable, because there was an outage of almost a day at one point because Entra ID was not available.

So they went and engineered and built this backup system for Entra. It runs in a different data center, it gets its power separately from the standard infrastructure, and it's geo-replicated everywhere. The resilience they brought in is that even if the main Entra ID goes down, that service can still issue tokens, the access and refresh tokens. New interactive auth will stop working, but anyone who is signed in stays signed in. If we're talking here now and someone pulls the plug on all the Entra ID servers, we won't get signed out. Behind the scenes, access and refresh tokens are renewed every hour, and if Entra weren't available, we'd all get signed out after an hour. This backup service makes sure that even if Entra ID goes down, traffic is routed to that service. It doesn't show a login page, but it keeps giving out tokens. It does some basic conditional access policy validation, but it keeps issuing tokens so that we never end up in a state where people start getting knocked out of Entra ID.

So there's a lot of engineering they did to build resilience into Entra ID itself at scale. If you search for Entra SLA, there's a page, and it shows up in your portal as well, that shows the availability, and it's 99.99%. So a lot of engineering goes into making it more resilient. But that's all Microsoft doing it at their scale. If you think back to your on-prem AD days, or even now, you're responsible for all of that. That part is no longer your thing to worry about, but there's a lot more that you do need to own and worry about as the Entra admin.


Craig Birch 18:37
And that actually goes back to that first thing. Microsoft put in a ton of resiliency, and it was the right thing to do. But it also gave us that feeling of, oh, we're safe, Microsoft's got us, because the service never goes down. That's not the case when it comes to other parts of it. Go ahead, Drew. I didn't mean to cut you off.


Drew Fuhrman 19:02
Oh, no, I didn't mean to derail the conversation. Craig, you make a really good point. Microsoft is meant to give you that warm and fuzzy feeling that you're always safe and secure with Microsoft. But when a lot of people hear the term recovery, they think about a catastrophic outage, where everything's down and everybody's scrambling. The reality is that a lot of the time, when you see a disruption, those incidents start smaller and quieter, kind of like what you were talking about, Merill.

So I guess the real question becomes: what are the most common self-inflicted identity incidents you see? Are they bulk changes? Can you talk to me about that a little bit, Craig?


Craig Birch 20:18
Yeah. I'm not going to say this happened in the real world, but maybe it did. Let's just say, hypothetically, that as an admin you're cleaning up some Entra ID applications. You've done your due diligence, or so you think. You put in the change control and you delete that application. Ninety days pass and no one screams, because what you find out is that the application is only used twice a year, but it's used in a business-critical part of the operation. Now what do you do?

We always think the reason we need recovery comes from the bad guy, but sometimes it doesn't. It really comes from day-to-day operations. Now you've deleted that Entra ID application, and it's bypassed the recycle bin. What do you do? You're in a pickle.

The other one, again not from experience, but I've heard of it: we go in and make a configuration change to a conditional access policy, or maybe delete one, because we think it's no longer in use. Then we find out that, guess what, it was in use, and now we have a huge security gap.

You mentioned bulk changes, and that's more users and groups. We've definitely automated a lot more. Automation is good, so don't take this out of context, but with automation we can make changes at a much faster pace than if we were doing one-off operations. So there are a lot of potential administrative and operational mistakes. And if you throw cyberattacks into play, now we're at a whole different level. What are your thoughts on that one, Merill?


Merill Fernando 22:31
Yes, and it can come from the most unexpected places. I know one true story that happened with Connect Sync. It syncs updates to your systems, and it has some built-in thresholds. What happened at one big org is that they did a huge reorg across the company and updated everyone's org data. Connect Sync has a threshold: when it makes more than a certain number of changes, it just stops, and you need to go and manually change that threshold, because you rarely update all the AD objects of all your users. So there are some safety mechanisms Microsoft built in, but as part of your business, you hit a point where you go, oh, we do need to do this.

So the org I know raised the thresholds way up to cover everyone. What they forgot was to bring them back down after the weekend when they made the change. And guess what happened? A few years later, someone made a change in Connect Sync that accidentally took out a whole bunch of people. It started deleting all the Entra ID objects for those users, and suddenly there was an outage, because they had disabled the safety net some years back. They had to get some work done, but they forgot to revert it to the safety limit. So it can come from anywhere. It could be accidental; it could just be an operational thing that happens.

The big bad thing we all talk about is attackers getting in and doing things to a tenant. In real life, that has happened a bit, but not so much, because with Graph there are all these different things. You can't just go and do something in minutes like you maybe could on-prem. Graph adds throttling, et cetera. But it is still a thing. Attackers can stay hidden in the cloud, but that's a different topic from the resilience we're talking about here.

Most of the stories I have come across are operational things, like a script doing something, or someone deleting a group. This was before we had recycle bins with a 30-day period. They deleted the group and had no idea who was in it, because that group had been built out over time, with people adding members to it. They had to walk back through the audit logs to piece together who was in the group and add those users back, especially because it was a very critical group used in a lot of different places. So yes, you have a 30-day window now, but what if you miss that 30-day window?


Craig Birch 25:44
Right, and the interesting one you mentioned is the delta sync and having that protection. I think it's either 300 or 500. I was like, man, did you read one of my support cases from back in the day? I used to do migrations, and I literally had that experience where we lifted the threshold, and then we ran a migration right after that. And guess what? We had an issue, and those objects disappeared from the tenant. We were able to recover, but it definitely feels like you were reading my mind on that one, because I lived it. So thanks for validating that.


Merill Fernando 26:12
Absolutely. And these are things we just forget. It's a case of unintended consequences, one thing leading to another to another, and that's when we end up with an outage. There are all these different things that are there to protect us, and they get switched off at different points in time, and it's that last straw that breaks the camel's back, and we end up with an outage.


Craig Birch 27:04
We're human, right? We all make mistakes. It's just the way it is.


Drew Fuhrman 27:11
Yeah. So I guess the question then becomes, what's the takeaway? Are there any best practices we can share with the audience about how to mitigate some of those risks without meticulously managing your environment? Any nuggets of knowledge you could share with us, Merill or Craig?


Craig Birch 27:36
Obviously, Microsoft has put some things in place recently. I want to talk about those first, and then maybe we'll unpack your question. Microsoft has introduced some Entra recovery capabilities, and we need these capabilities. I want Merill to explain them, because he's been living in this space for a very long time. If he can explain what those new capabilities are, then I want to talk through some of the other things we might need to look at, and that will answer your question, Drew, if you don't mind.


Merill Fernando 28:15
Yes. Interestingly, Microsoft recently introduced Entra backup and recovery. We've had Microsoft 365 backup and recovery on the M365 side of the house, and the Entra team just added this. For those who may not know, there was an open-source project called Microsoft365DSC, Desired State Configuration, and they basically took that and made it a hosted service. So it's sort of powered by Microsoft365DSC, but there's a bit more happening, and they now provide it as a hosted service in Microsoft. You can still run Microsoft365DSC yourself if you want. It's not as polished as what Microsoft is doing, but you don't have those service limits and other things, so if you're just looking for the basics, you can do that.

There's also an open-source project I worked on when I was at Microsoft called EntraExporter. It's not a recovery tool. We made it as just, OK, I want a backup. It exports everything to JSON. It doesn't recover things, but you at least have a text file with the list of groups, and you can figure out what to do with that in a very bad emergency.

So there are those open-source projects, and you now have Entra backup and recovery built into Entra. It's very basic at the moment. There are certain limits on how many objects and which objects it supports out of the box. Microsoft has just started investing in and building out some of these things.

But it's also really important to remember that they're not bulletproof. If an application or service principal gets deleted, it's past 30 days, and you restore it, the actual object IDs change. So if you have code that looks up certain IDs, those scripts and other applications will still fail if they rely on the object IDs being the same, because when you recover, they have a different object ID. You just need to be aware of some of those things. It's the nature of how Entra was built: you can never get the same GUID object ID back when you recover, so there's nothing that can be done there. A lot of it is more about ISVs and others building apps that are more resilient and don't rely on brittle things like that.

So there have been some different investments Microsoft has been making in this area. They're very, very new, a V1 Microsoft thing. We used to say V3 is when it's a mature product at Microsoft. I'd say backup and recovery is somewhere around there at the moment.


Craig Birch 31:32
And you mentioned some limitations. I think it keeps five days, if I'm correct. It also has limitations on the objects you're able to recover. Hard-deleted objects weren't something it could handle, and there was no hybrid. I believe there was a limitation around hybrid, at least when it was first announced. I don't know if that's still the case, and that's why I'm asking.

When we talk about hybrid, traditionally I think of, and I'm going to date myself again, AAD Sync synchronizing to Entra ID. But more recently, as I was thinking about what hybrid really means, it's not really that anymore, because hybrid extends way beyond our on-prem Active Directory syncing to Entra ID. With Entra ID being able to do so many different things and reach so many different systems, Entra ID itself has been a hybrid identity platform from the get-go. So what's your thought process when people say hybrid? Should we still think traditionally, that it's just AAD Sync, or should we expand our mindset and think broader?


Merill Fernando 32:57
Yeah, when you think of cloud backup, the very first thing people say is, oh, we're syncing things from on-prem. So if someone messes up these objects, we can just turn on sync again and it will write things back and restore those objects. If someone messes up all of the user info and so on, hybrid definitely helps that way. That's assuming your on-prem backups were all done, which many orgs did. You had various products, and you did the backup because you were responsible for a lot of it.

But as we've been talking about, the whole Entra IdP is more of a service now. The backup of just the user objects, and maybe the devices, is now maybe 10% of it. When you first started, that might have been 90% of your Entra estate. But as you start using more features in Entra, as users create more cloud groups like Teams groups and security groups, if you're using identity governance with access packages and some of those Entra features, and definitely as you put in conditional access policies, a lot of things are no longer hybrid. The hybrid slice of Entra is now maybe 10%, a much smaller slice of the things you're responsible for, like all the apps and service principals. Those all become an important part of it.

And the practices we used to have are still important. Take that mining company I told you about, where they deleted the service principals. The only way they could recover was to go back and re-add every app: every Salesforce configuration, every Workday, ServiceNow, all of those things. They had to redo them. But because they were an enterprise, they were very diligent with all their CABs, access changes, and advisory board reviews, so they had documents. Not perfect, but they at least had runbooks for every app they onboarded. So they just had to go through them. It took them months, but they did the more critical ones first, ran through each of those things again, and got it recovered. They still had a paper trail they could follow to get it back.

That sort of operational resilience, which we feel is just writing documents no one ever reads, is more like an insurance policy you can go back to. And now with AI, I feel those are the places where we can actually use AI a lot more, to document these things a lot faster, because you don't need a human being to do those things. As you make config changes to Entra, if you put it into practice to script them out and use AI agents to write and deploy those scripts, you start building that operational resiliency as well, and the cost of those things is going down. There are lots of backup tools and solutions that are really good and will help get you back faster. And having these AI agents gives you the confidence to ask, are we good now? Is it in a workable state?


Craig Birch 37:00
I have a question for you. Entra ID has been around forever, and we're only now getting Entra ID backup and recovery. What's changed? What's really driving this new momentum at Microsoft? I have some ideas, but I want to get your thoughts, and then I'll come back with mine.


Merill Fernando 37:26
I'd say Microsoft is very metrics-driven, especially when it comes to justifying building a new feature. Backup and recovery has been one of the top five asks from customers. Every year they do surveys and go through what they're going to build next year, and it always kept coming back in the top five: we need some sort of native capability to back up and recover. So it's always been there as an ask from customers, but they had a lot of other things to build out.

And when Microsoft builds something, they think about it at global scale. That's why the first part of the story was things like recycle bins. They needed things that work for everyone. It's usually the enterprises and bigger orgs that plan up front and have a whole backup and recovery strategy. When you're more mature, you do that. But you have to remember Microsoft has millions of Entra ID tenants, and a lot of them might be small businesses with just five people who use it for Exchange and email.

But they need the same things. The recycle bin covered a lot of those, the 80% of cases where something bad happened, they noticed it, and they can now self-recover. That covers about 80% of the scenarios Microsoft sees across the board, across its 8 million customers. So the need was always there. When Microsoft does things, you might think it doesn't make sense: why are they doing just these simple things? But they're very heavily metrics-dependent, and the scale they look at is very different from the scale you and I look at as an enterprise or one Entra org. In their view, it's people who don't know anything about Entra. How do we help them? They deleted the group; how do we help them recover? That's where features like the recycle bin come in, and they're useful. You might say, that's just 30 days, what can I do with that? But that covered 80% of the support tickets Microsoft was getting.

So that's been the history. Over time, they started building out other capabilities to cater to scenarios that would cover the 80%. But I'm sure a lot of the folks who joined this call today are in that 20% or 10% with big orgs who can't leave it to chance or risk not having resiliency in their Entra infrastructure.


Craig Birch 40:35
You made some valid points. It's been requested by a lot of organizations, and Microsoft obviously supports way more than just enterprises. They support everything from the smallest customer to the largest. When I was thinking about why now, two things came to mind, and they weren't those two, so I appreciate that context.

The first one is that we're dealing with a new world of modern-day cyberattacks, we're seeing rapid change, and Entra is an attack surface. That's number one. But that wasn't the only thing I was thinking about. We're moving to an agentic world. We're going to see rapid change internally because of agents making changes.

So is Microsoft now looking at it as: this is no longer just the most requested feature from our customers; it's a must-have. It's table stakes now, and we need to get at least the users, the groups, and maybe some of the core pieces covered initially? Maybe you have some inside track on whether this is going to expand into other areas, but it's obviously not covering some of the bigger areas I would have expected from Entra ID backup and recovery. That's my personal opinion. Any thoughts on that?


Merill Fernando 42:20
Yes, I think that's been the nature of it. It used to be just syncing objects, but the estate keeps growing. Entra itself went from Entra ID to the Entra Suite, so now you have a lot more services, and now you have network components from Global Secure Access being added. Your dependency as an org keeps increasing, and a lot of those things live only in the cloud. There are all these other things built on top, and yes, now we have agents as well. A lot of these are purely cloud-only objects and services.

So backup and recovery becomes really important, because whether it's intentional or unintentional, you can cause a lot of issues and outages just by messing up things that run only in the cloud. And yes, they were certainly seeing a lot of issues and support tickets come through, and that's what they've been working on: adding those capabilities.


Drew Fuhrman 43:39
Sorry, I was just looking at the clock, and we've got about 15 minutes left. I want to keep the train moving, because I do want to give anybody in the audience a chance to ask questions. If you're still with us, drop any questions in the chat so we can address them toward the end as we start to wrap up.

Craig, you made a really interesting point: this is table stakes now. We have to make sure we have access to all of this data in case we do go through an outage. So let's move from the topic of here's what can happen to here's how we can plan and mitigate some of these challenges as an organization. Craig, I want to start with you. If a team wanted to improve their identity recovery readiness in the next 30 to 60 days, where should they start, in your opinion? And then Merill, I'll follow up with you right after with a similar question.


Craig Birch 44:50
The first thing I think the business needs to understand is what applications and business dependencies are riding on the Entra ID identity fabric. I would start there.

The other one is something a lot of organizations don't do. We put things on paper, but we rarely ever test. We never test a recovery incident. And it's not just Entra ID; we've had this for years with Active Directory recovery as well. It's hard to test these things, but we need to put things in place where we're testing real-life recovery scenarios. Pick two or three recovery scenarios, test them, and make sure you're able to recover from them.

And the other one is: if we can't get to Entra ID, we can't recover. So Merill, how do we protect the identity recovery path? What are some good practices organizations should follow around privileged roles and making sure they're not locked out of their tenant? You can't recover if you can't get to your tenant. Can you give some insight there?


Merill Fernando 46:18
Yes, absolutely, and I'll share this link. Security and recoverability go hand in hand. I was part of the team that built aka.ms/entra/security. It's basically Zero Trust practices, and a lot of the questions you're asking are covered by them, because while you're building to keep your tenant secure, that builds a lot of native resilience and recoverability into it. You're doing things like creating your emergency access accounts, keeping attackers at bay by following better practices, making sure your apps don't use secrets, all the different things that could otherwise lead to an attacker coming in and doing things in your tenant. And it covers conditional access best practices and things around that.

The big thing, from my own experience: I started my own SaaS company, and I was counting yesterday. I already have eight Entra tenants. I'm a one-person SaaS, but I'm following the practices Microsoft has in the docs. For my one production Entra tenant, I have a pre-production tenant and a dev tenant, so that's three tenants. I have Entra External ID for external users signing in to my apps, so that's another tenant. When you add my corporate tenants and my external tenants, that's three and three, six. And because I'm building consumer apps, I have another Entra tenant, separate from those, where my multi-tenant apps live. So they're all separate and resilient.

Yes, it's a bit more work, but you need to plan some of these architectural things. If you do it from day one, it's easy. If you didn't, you should start today. Is your Entra tenant the only production tenant you have? You should put in the practice of having at least one non-prod or pre-production tenant, and ideally a dev tenant where more people can work. The whole architecture and resiliency piece is easy to start from day one and really hard to retrofit, but you should do it. It should be a practice, and you should try to get those things in place.

These aren't technical things. They're general best practices we have for every system you onboard. You have prod and pre-prod, you test things in one place before doing them in another, and for your critical conditional access policies and critical groups, you should have a replica tenant where you test them before you roll them into prod. The bigger part is making sure you maintain it over time. Not every CA policy you change, but the core part of your functionality, your core conditional access and authentication method policies, should be kept in sync across those environments. These are the more mature practices.

There are, as I said, millions of tenants, and it's these mature best practices that will help you in the long run. Microsoft has excellent documentation on everything from resilience to recoverability. It's just that very few people read all the docs, let alone follow them. So it's definitely something we as Entra practitioners need to do: step up and think about the bigger picture in these scenarios.


Craig Birch 50:48
You're mentioning multiple tenants. Large enterprises might have a dev and a prod tenant, but that might not be the case for a small business. They might have one production tenant. To be honest, they might not even be managing that tenant; somebody else could be managing it for them. You're talking about testing here and then moving it there, but if something happens, you're relying on internal knowledge, on Entra ID backup and recovery, and on its limitations.

If I go back to saying we need a recovery plan, what does that plan look like? More importantly, if I'm a CISO, how am I measuring it? How do I measure success in recovery when it comes to identity? What's the magic framework there? I don't know if one exists.


Merill Fernando 51:54
No, I don't think so. That's why you need all that maturity, or at least some tooling. I think Cayosoft has some really good tooling as well, where you can say, let me just rely on that and let someone else take care of this for me.


Drew Fuhrman 52:18
Oh, you saw my shameless plug in the chat there, Merill.


Craig Birch 52:23
Sure, Cayosoft has excellent tooling around recovery. But even with tooling, there has to be a way to measure it. How do I test it? It's beyond just tooling, so I don't want to make it about that. I didn't know if you had some insight from your experience.


Craig Birch 52:44
Maybe this is a different, longer-term conversation. I know we're almost out of time, and I think we want to open it up for questions from the audience. Are there any?


Drew Fuhrman 52:53
I'm not seeing any questions from the audience in the chat right now. If anybody has questions, please drop them in; we still have a little time left. I have a question, though, about those solutions and making a plan for recovery. Are there any assumptions you see a lot of people making that might be dangerous? When you're coming up with a recovery plan, you might make a couple of assumptions. Is there anything we want to call out to everybody on the call today to be on the lookout for?


Craig Birch 53:22
The first one is that I can recover. I'm going to go with the most obvious one here, guys. The first mistake is assuming we can recover, and that's not the case in a lot of organizations, because they've never tried to recover, so they don't even know what they can recover.

The second is: if I can recover, what can I recover? Are there limitations to my recovery plan that I'm not aware of? I'm assuming I can recover users and groups and that gets me far enough, but Merill made some really good points. It's no longer just users and groups. There's a whole ecosystem that goes with Entra ID. If I'm assuming users and groups are good enough, that's probably not the case in modern organizations. We have to account for conditional access policies, Entra ID applications, and beyond. I can't name everything, but there's a lot.

So don't assume you can recover. Even if you have recovery, you need to understand its limitations. And then I want to know how you measure it. This is where organizations really need to dig in and ask: how do I measure success? What does recovery look like? It's not going to be the same for every organization, but there are probably some fundamentals we can build. And it sounds like there's nothing out there that helps them understand what that is. Maybe this is an opportunity.


Merill Fernando 54:40
Yeah, I definitely think so, because what it means for different orgs is very different. Some of the enterprises I work with define business-critical applications: everything else can go down, but for our business to survive, these N apps have to be up and running, or the business doesn't exist anymore. They define those, and then you need to work back from there and make sure you can test them, and make sure everything in that critical path, like the security groups used by those apps, is there. That's where you need to be really sure.

You can assume it's fine if your HR system isn't working, but the thing that's printing money can never go down; that's the business-critical part. For a consulting firm or someone else, though, the HR system might be the business-critical thing. So it changes depending on the org, and they need to work on that. You can't say all 100 or 200 or 1,000 apps your org uses are business-critical and put in all the resourcing to make sure they're all up and available. You focus on what's really important.

And that's where test environments come in. Even if you buy different products and suites, you need to test them somewhere. Is the recoverability actually working? When things don't work out, are you going to wait to test it in production when something goes wrong?


Craig Birch 56:55
Yeah, I agree. I mentioned earlier that we need to map critical identity dependencies to the business. What you're really talking about is minimal viable business: what's really required to run my business? There's no magic bullet for that, because, as you mentioned, it depends on the business and what it does. You need to understand what your minimal viable business recovery plan is and build around that.


Drew Fuhrman 57:25
We did have a question come in from David, if you're both OK with that, and then we'll start wrapping up because we're running out of time. David asks: given the potential impact Entra ID outages pose to business operations, what are some of the measures companies should have in place if they want to implement SSO? What do you think?


Merill Fernando 57:52
It's a good question. This comes up quite a lot. Companies say, OK, we have Entra, but we have these other IdPs, maybe ADFS, and now we want to move to Entra as our strategic, main one. Then they ask, what happens when Entra is down? It's a very visible question people can ask. But if you look at Entra resilience, which we'll share, like the stories I told you about the backup service, an Entra outage alone hasn't happened in a long time, because of a lot of things they do, like the cell-based architecture. They've reduced the chance of a global Entra outage ever happening, though there could be a localized one. The chance of that happening is a lot lower than all the other things we spoke about: deletes, unintended deletes, and various other things that can happen.

This usually comes up because it's a big question, but I've never seen it in real life, because usually if there's an Entra outage, there's also some other outage. Teams, Exchange, or your internet, everything else is down too, so Entra being up at that time wouldn't make any difference.

On the other side of the story, I do know orgs that, as part of regulation, have to have a backup in place. They've set up an on-prem IdP that they manage, assuming everything goes down. In Australia, for example, we have critical industries, like water suppliers, power plants, and other critical infrastructure. They have alternate scenarios where they have an on-prem IdP, and they've done the paperwork and the practice so that, for an application that's also on-prem, if the whole internet goes out, they can switch the application to use the backup IdP. But it's still an outage. You're not going to be up within minutes. It might take an hour to reconfigure the app to use an IdP that's not Entra.

And you can't plan that for all the hundreds of apps you're adding, because it means testing every app with two identity providers and what happens when you swap users between them. It's a disaster recovery scenario at that point. It's not worth it for everyone, but some orgs have to do it, and those are the extreme scenarios where that sort of thing makes sense.


Craig Birch 1:00:42
Yeah, and you mentioned it: a whole Entra or Microsoft outage is rare. Maybe it's happened a couple of times. But a tenant outage due to an insider oops or an outsider oops at the tenant level is more likely to happen. I think that's where we need to put the focus: what can we do as an organization to make sure our tenant, not the Microsoft fabric, which is Microsoft's responsibility, is resilient? We have recovery, we know what our recovery plan is, we know it's tested, and we know what business operations run on it. That's how I would close.


Drew Fuhrman 1:01:29
Yeah, and I think we're at time now. Thanks for the question, David. Craig, Merill, thank you for joining us. As we wrap up, I want to reiterate the key message: identity recovery is no longer a narrow directory administration topic. It's part of M365 availability, security, continuity, and the business resilience we've talked about today. Entra ID, Active Directory, synchronization, access policies, roles, applications, and devices are all connected, and the practical challenge is making sure recovery plans reflect that reality.

Before we go, Craig and Merill, first, thank you again. Second, yes, this call is being recorded, and we'll be sending it out. That gets asked all the time. And finally, do you have any closing thoughts you want to share quickly? Merill, we'll start with you.


Merill Fernando 1:02:41
Oh yeah. I'm very happy to see so many folks join in to talk about Entra recoverability and resilience. It's often left aside. It's not a fancy subject, but it's a reality for a lot of Entra admins, so it's good to see everyone join today. And for those who aren't following my newsletter, feel free to sign up. Thanks for sharing the link. It's all the weekly Entra news, coming to your inbox.


Drew Fuhrman 1:03:14
Great. What about you, Craig? Final thoughts for the group?


Craig Birch 1:03:18
Well, I think I covered most of them. Obviously, Cayosoft is here to help you with your identity resilience, and we have solutions to help you with Entra backup and recovery that go beyond what Microsoft is offering from a backup and recovery perspective. But again: test, validate, and measure that you're able to recover. And more importantly, that minimal viable business. Start there. If you don't know what your minimal viable business is, you're in trouble, so you need to get in front of that. Thanks for joining.


Drew Fuhrman 1:03:55
Yeah. And thank you both again for your time, and thank you to everybody who joined today. I know we're all busy with a lot going on, so it's really nice to see everybody here. We're going to wrap this session up and call it a day. I hope everybody enjoys the rest of your Tuesday, and happy September, all. Take care.


Craig Birch 1:04:14
Thanks for waking up for us, Merill.


Merill Fernando 1:04:16
Oh, good to be here. It's a fun topic. See you, folks.


Craig Birch 1:04:19
All right, later.


Drew Fuhrman 1:04:22
All right, see ya.

Recorded September 1, 2026

Time: 1:00 PM ET / 12:00 PM CT

Format: Live Webinar

As organizations move deeper into Microsoft 365, identity has become the operational foundation of the business. Access to email, collaboration, applications, devices, and administration all depend on Entra ID. At the same time, many organizations still rely on Active Directory, synchronization services, and hybrid identity configurations that can introduce operational risk and recovery challenges.

The conversation would touch on areas such as:

  • Why Microsoft 365 availability depends on identity
  • How Entra ID impacts users, applications, devices, and administration
  • How administrative mistakes, synchronization failures, and configuration changes can disrupt operations
  • Common identity recovery scenarios organizations often overlook
  • Microsoft's native backup and recovery capabilities and restoring identity services, configurations, access, and operational continuity
  • Practical approaches for improving resilience across Entra ID and hybrid identity environments

Your Presenters:

Craig Birch

Principal Security Engineer - Cayosoft

Craig brings over 24 years of hands-on experience in Active Directory architecture, operations, and security. Craig’s extensive expertise spans Active Directory, AD migrations, identity management, Entra ID, and M365 technologies, with a strong emphasis on securing these environments.

 

Craig_Headshot_For_Website

Merill Fernando

Founder & Identity Architect- Jozra

Merill is a Microsoft-focused identity and security expert known for helping IT and security professionals better understand Microsoft Entra, Azure, and modern identity management. Through his public content and community work, he shares practical guidance, tools, and insights that make complex Microsoft identity topics easier to apply in real-world environments.

Merill Fernando LP Headshot
© 2025 Cayosoft, Inc.