Speaker 1: In 2018, Microsoft bought a small company called FSlogix. What that company built became the default way virtual desktops remember their users, and it now ships inside licenses that a lot of you already own. The idea is simpler than the name suggests, since it keeps a whole Windows profile in one file and hands that file to the user wherever they sign in. We will go through how that works, where the file should live, and the mistakes that cause most of the trouble. By the end, you will know whether this belongs in your environment. A user profile is everything that makes a desktop feel like yours. It's the wallpaper, the browser bookmarks, the way Outlook is set up, the apps you pinned, and the files sitting on your desktop. On a laptop, that profile lives on the disk and just stays there, so you never have to think about it. Virtual desktops change the math. In a pool, host pool, desktops are shared and built to be replaced. A user might land on a different session host every time they sign in. Monday, it's host one, and Tuesday, it's host four. The user doesn't know that, and honestly, they shouldn't have to. The result is a desktop that forgets them. The wallpaper resets to a stock image. Outlook starts pulling the mailbox down from scratch, and the shortcuts they made last week are gone. Sometimes Teams asks them to sign in again, and OneDrive starts a fresh sync like it's the very first time. They file a ticket, and you go hunting for a bug that isn't there. Windows assumes the profile belongs to the machine it's running on. When that machine is temporary, the profile needs somewhere to live that outlasts it. That's the whole problem in one sentence. So the fix isn't to copy the profile faster, it's to stop copying it at all. FSLogix is Microsoft's profile management tool for virtual desktops. Microsoft acquired the company back in 2018 and folded the technology into its own stack. The profile container is the core piece. It takes the whole user profile and stores it inside a single VHD or VHDX file on a network share. If those two file types are new to you, think of them as a virtual hard disk. It's the same kind of file a virtual machine uses for its own disk, and VHDX is the newer format that you want. At sign-in, the container gets attached to the session, and Windows sees a normal user profile. Microsoft's own wording is that the profile appears in the system exactly like a native one. There's no copy at sign-in and no copy at sign-out. That last part is where the speed comes from. Older roaming profiles copied files back and forth across the network, and the bigger the profile grew, the slower the login got. FSLogix mounts a disk instead, so the profile is ready the moment the container attaches. It also captures more of the profile. The container holds nearly everything under the user's profile folder, including the app data, local data that older roaming profiles skipped. That's the folder a lot of applications quietly store their settings in, which is why your user's app state actually survives the move to another host. Two more details come up often. The container is created as a dynamic disk by default, which means it starts small and grows as the user writes data, and it's a single file which makes it easy to back up and easy to move between storage locations. The container is tied to the user, not to any single machine. Once it's sitting on the share, it follows that person to whichever host they land on, in whichever region as long as the host can reach the storage. That's the property that makes a shared desktop feel personal. A couple of things to note. Because the container is a file on a share, the network path to that share matters more than almost anything else in the design. If the path is slow, the desktop is slow. That's the trade-off you accept for keeping the profile off the machine. Let's walk through a sign-in step by step because the moving parts explain the mistakes later. The session host reaches out to the file share, finds the container that belongs to that user and mounts it. FSLogix then sets up a set of redirections which point the profile paths at the mounted container so everything lands in the right place. Some things are deliberately kept out. Credentials, stores and the office file cache get redirected to a local folder on the session host instead of the container and that folder is deleted at sign out. Those items shouldn't roam between machines so FSLogix doesn't let them. This is also why the profile doesn't carry a pile of secrets from one host to the next. Other items like temporary app data and device identity data get stripped out of the container at sign out on purpose. The product teams that own that data recommend it because roamed identity data causes more problems than it solves if you need to keep something else out. There's an XML file for that. You list the paths you want excluded or redirected and you deploy the file centrally so every session host gets the same rules. A common shortlist is browser cache, the Outlook data file, and the OneDrive sync cache, because those grow fast and don't really need to travel. Here's a limit worth knowing early. Adding a path to that XML file does not clean up data that's already inside existing profiles. The exclusion applies going forward. If a profile already has the data, you either rebuild the profile or delete the content from the container yourself, and both carry some risk of data loss. And if the storage can't be reached at sign-in, the user doesn't get their profile at all. They land on a temporary profile, which looks like a fresh desktop, and quietly throws away anything they do inside it. That symptom is worth recognizing early because it almost always points back to storage or permissions. At sign-out, the container is unmounted and written back to the share, and the next host can pick it up. Three names come up together whenever people talk about FSLogix, and they trip people up. Profile Container, Office Container, and Cloud Cache. The Profile Container does most of the work, and it already includes the Office data. For most people, that single container is all you need, and Microsoft recommends the single container setup. The Office Container, sometimes called ODFC, is a separate container just for Microsoft Office data, things like the Outlook data file and the OneDrive cache. It makes sense mainly if you already run another profile tool and only want FSLogix for Office. It can also help if you want Office data in a differently sized container than the rest of the profile. If neither of those applies to you, skip it and keep things simple. Cloud Cache isn't a container at all. It's a setting you can turn on. Instead of writing straight to the network share, the session writes to a local copy first and syncs to storage in the background. You can point it at more than one storage location and if the network drops or one location goes offline, the user keeps working from the local copy. A couple of things to note about Cloud Cache, because it isn't free, it needs fast local disk on the session host, and it adds a little time to sign in. On non-persistent hosts that get rebuilt regularly, that local copy is wiped every time, so every sign in rehydrates from remote storage anyway, which mostly defeats the point. In a single region setup that's already on fast redundant storage, Cloud Cache often adds complexity without adding much you didn't already have. There's one hard edge worth stating. If Cloud Cache can't read a piece of the profile from any storage location, the session can hang or worse. It's a resiliency feature, not a backup, and it doesn't replace a restore plan. This is the part that decides whether your users have a good day, so it's worth getting right. The container sits on SMB storage. On Azure, the two common choices are Azure Files and Azure NetApp Files. Microsoft recommends Azure Files for most customers, and NetApp Files is the step up for heavier or larger environments. Keep the storage in the same region as the session hosts. Distance adds latency, and latency on profile storage is felt directly as a slow sign-in. Microsoft's own guidance is blunt about this, because how close the storage is to the desktops decides how the desktop feels. Now, for the number that matters, which is IOPS, or input and output operations per second. A practical rule is about 10 IOPS per user at steady state and closer to 50 per user during sign-in and sign-out. Here's what that looks like in real life. For 100 users, that's roughly 1,000 IOPS through the day, but it can spike toward 5,000 in the window where everyone arrives at once. If you size for the average, the first 20 minutes of every morning will be the worst part of your user's day. Size for the burst instead. Azure Files comes in two tiers. Standard runs on spinning disks and is fine for small light use. Premium runs on solid-state drives and is the usual pick once you pass a couple hundred light users or run anything heavier. The price follows the performance and standard also bills per transaction, which makes its speed very under load. For something every user touches to log in, that variability is exactly what you're trying to avoid. One more practical note. Whatever you choose, set up monitoring and alert yourself when the share has less than 20% free space because a full share causes the same symptoms as a broken one. Cost is worth a thought here because this is usually where the money goes. Profile storage grows with your user count and with how much each profile carries, so it's the line item that rewards good hygiene. Cleaning up profile bloat isn't just tidiness, it's a direct cost saving. Hold on to that because, uh... Storage is where most FSLogix deployments get into trouble, and that's where we're headed next. Licensing first, because the good news here is that most of you already have it. FSLogix isn't sold on its own. It's included with Microsoft 365 E3 and E5, with A3 and A5 for education, with F1 and F3, with Microsoft 365 Business, and with Windows 10 Enterprise E3 and E5. It also comes with Windows 10 Education A3 and A5, RDS client and subscriber access licenses, and the Azure Virtual Desktop per user access license. In plain terms, if you run Azure Virtual Desktop or Windows Enterprise desktops, it's likely already in your licensing The work is configuration, not procurement. Beyond licensing, you need SMB storage with permissions set up correctly, an identity setup that lets your users reach that storage, and the FSLogix agent installed on the session hosts. The agent is a small install, but it has to be baked into your golden image or added by your deployment process, so it lands on every host a user might reach. If a host is missing it, the user gets a local profile on that host and loses their settings there, which is a frustrating bug to chase. Your identity setup matters here too. Whether that's on-premises, Active Directory, Entra Domain Services, or Entra Kerberos, the point is that users need a clean path to the file share because if authentication fails, the profile doesn't load at all. It runs on Windows 10 and Windows 11 and on Windows Server 2016 through 2025. It also runs outside Azure. It works with Citrix, Omnisa Horizon, Parallels RS, and plain remote desktop services, and it runs on-premises, hybrid, or multi-cloud, as long as the storage is close to the desktops. One honest limit before we move on. Included in your license does not mean zero setup. Microsoft doesn't design your storage for you, and it says so plainly, that part is on you, so let's look at where these deployments usually go wrong. This is the section I'd want someone to walk me through before my first rollout, so let's go through it. Permissions are the big one. Get the share permissions wrong, and profiles fail in ways that look like something else, so you end up chasing the wrong problem for hours. The pattern that works is fairly simple. Give your user group modifier access, give administrators full control, and set the creator owner so each user's own container inherits the right access. Keep admin accounts out of FSLogic's profiles so you always have a clean way in when something breaks. Antivirus exclusions are the next classic. The exclusion isn't limited to the container file. The small sidecar files sitting next to it, the log and metadata files need the same exclusion or the antivirus logs them and you get sharing violations and temporary profiles. If 20 users all log in at once and every container gets scanned, a single host can fall over under the load. Sizing is the quiet one. The default container size is about 30 gigabytes and the setting is effectively one-way. Raise it and existing containers grow at next sign-in, but lowering it does nothing to containers that already exist. Profiles also don't shrink on their own. When a user deletes files, the space is freed inside the container, but the file on the share stays the same size. Compaction runs at sign-out and can claw that space back, but it only runs if at least 20% of the used space is recoverable and it quietly stops working if the Windows Drive optimization service is disabled in your image. There's no error. and it just does nothing. So 30 gigabytes is a starting point, not a rule. A light user with a small mailbox and everything else in one drive might genuinely need 10 to 15 gigabytes, while someone with a large mailbox and heavy local apps can pass 60 gigabytes without anything unusual happening. Measure real profiles on your share before you commit to a number and a size for the biggest real profile you have plus some headroom. Slow storage is the most common cause of FSLogic's trouble overall. When the storage can't keep up, users get slow logins and sluggish apps, and they blame the desktop, not the storage. Corruption is the one that ruins a day. An improper shutdown or a storage interruption can leave a container that won't mount, and without a backup, you're rebuilding that user's profile from scratch. Treat the container as something worth protecting, not just something that exists. Before you go live, test the login load rather than just the configuration. Push a realistic number of users through a sign-in at the same time because a storage setup that feels fine with a handful of test accounts can behave very differently when 100 people arrive at 9 in the morning. One trap to avoid because it looks helpful, there's a setting that appears to fix sign-in quirks by roaming identity data. Don't turn it on for cloud-joined or managed devices. It ties identity to the wrong thing and causes worse authentication problems than the ones you were trying to fix. After all of that, it's fair to ask whether every desktop actually needs this. The honest answer is no. FSLogix is built for the case where a person could end up on a different machine each time. That's the pooled, non-persistent host pool, and that's where the roaming problem actually exists. Windows 365 Cloud PCs don't use FSLogix profile containers necessarily. A cloud PC is persistent and assigned one-to-one, so the profile simply stays on the cloud PC between sessions the same way it would on a laptop. There's nothing to roam, so there's nothing for FSLogix to fix. AVD personal host pools work the same way. One session host is assigned to one user, so a roamed profile is convenience and portability rather than a requirement. It can still be worth having for backup and recovery, but the architecture doesn't force your hand. Here's a concrete example. If every user always lands on their own dedicated desktop, you aren't fixing a roaming problem, and adding profile containers mainly adds one more thing to maintain. On a single persistent desktop, FSLogix is optional. One current caveat worth knowing. An April 2026 Windows Server update changes the default Kerberos encryption type from RC4 to AES-SHA1. File shares that host FSLogix containers and aren't upgraded to AES-SHA1 can hit access problems after that update, so it's worth checking before the update lands, especially if your profiles sit on an on-premises share or on Azure files using Active Directory authentication. Doing this by hand through group policy and the registry is fine for one host pool. The settings are documented and you can get a single pool working in an afternoon. The tricky part isn't the first host pool, it's the 20th. It gets painful across a growing estate because the same settings, permissions, and exclusions have to stay consistent every time a new host pool appears. One host pool with a slightly different storage path or a missing exclusion is enough to generate a fresh batch of tickets. That's where automation earns its place. Nerdio Manager, for example, applies FSLogic's configuration as policy at the host pool level handles the storage account permissions an agent set up in a guided sequence, and watches capacity so you get warned before a share fills up. It also ships a scripted action to shrink profiles offline when per sign out compaction can't claw the space back. The honest framing is that automation removes the repetitive manual work. You still make the design decisions about storage, sizing, and exclusions, and those decisions are the ones that matter most. So here's the short version. Keep the profile in one container that follows the user. Put that container on storage that's close and fast enough for the morning rush. Set permissions and antivirus exclusions correctly. Size for real profiles rather than a round number. And leave it off desktops that don't roam. Here's the question I'll leave you with. Where does your profile storage live today, and how are you measuring whether it keeps up at the busiest time of day? If this made the topic feel less intimidating, Subscribe to M365FM to master the tools and the mindset shaping the future of work.