Ignite 2018 will be remembered for a lot of Azure noise, but the announcement that made our VDI practice sit up was Windows Virtual Desktop: multi-session Windows 10 in Azure, Microsoft-managed control plane pieces, and a licensing story that could rearrange the economics of desktop virtualization for anyone already paying for the right Microsoft 365 or Windows entitlements.
This is not a product review of something we have run in production — preview is not even open as we write. It is a field read: why the announcement matters if you already run Citrix XenDesktop, RDS 2012 R2 session hosts, or are mid-plan on the next VDI hardware refresh.
What was actually announced (as we understand it today)
Strip the keynote gloss. The pieces that matter to architects:
- Multi-session Windows 10 in Azure. That sentence alone is the plot twist. For years, dense session-host economics lived on Windows Server RDSH; Windows 10 was one user per OS instance unless you paid Citrix/VMware taxes on top of complex stacks. Multi-session client OS in Azure changes the density conversation.
- A Microsoft-managed service plane for brokering and related control functions — less of the "build your own connection broker farm" toil, at least in the story Microsoft is telling.
- FSLogix in the backdrop. Microsoft bought FSLogix earlier this year; profile containers are the missing piece every VDI architect names when multi-session gets real. We assume — and will verify in preview — that FSLogix is first-class in the WVD world.
- Licensing entitlement tied to Microsoft 365 and Windows entitlements many enterprises already own. If that holds under the final product terms, the "second Windows tax" problem that made pure VDI expensive gets partially dismantled.
Everything else is speculation until we put hands on preview bits.
Why Citrix and RDS estates should care without panicking
We still deploy XenDesktop 7.15 LTSR and RDS 2012 R2 + App-V for good reasons: mature HA, HDX over hostile networks, app layering ecosystems, and customer ops teams who know the stack. WVD does not obsolete those overnight.
What it does is change the default recommendation for greenfield and for Azure-bound desktop strategies:
- If the driver is "get desktops into Azure with Windows 10 compatibility and Office," WVD will be on the short list the moment preview is usable.
- If the driver is "advanced HDX, complex published apps, existing Citrix investment," Citrix on Azure (or hybrid) remains rational.
- If the driver is "cheap session hosts on Server OS," RDS still works — but multi-session Win10 may steal that narrative for user experience reasons alone.
Our advice to clients this week: do not freeze Citrix refresh projects already in flight. Do put a WVD pilot on the 2019 roadmap and stop buying five-year on-prem VDI hardware on the assumption nothing in Azure will be viable.
The questions we will answer in preview
When bits drop, our lab will focus on boring operational questions, not demos:
- Identity. Azure AD join vs hybrid join vs traditional domain join for session hosts. Which identity path is required for FSLogix and Office.
- Image management. How painful is golden image update compared to MCS/PVS habits.
- Profile performance. FSLogix container on Azure Files vs other storage; login duration under concurrent load.
- Network. Path from user to host pool; when you still need a gateway story like NetScaler or RD Gateway.
- Cost model. List prices mean little; we will model users-per-vCPU for task, knowledge, and power workloads and compare to a known XenDesktop Azure deployment.
- Management plane gaps. What still needs ConfigMgr, Intune, or third-party tools on day one of GA.
Until those are answered with telemetry, anyone selling "WVD replaces all Citrix next quarter" is selling.
Licensing: the entitlement bombshell, carefully
The strategic play is obvious: Microsoft wants Windows and Office consumption in Azure under subscriptions customers already pay. For shops drowning in VDA and third-party connection broker costs, that is attractive.
Caveats we are already writing into client emails:
- Entitlement matrices change between announce and GA. Design options; do not sign five-year contracts based on a slide.
- "Included" never includes storage, network egress, or the ops labor to run images.
- Bring-your-own Citrix or other partner layers may still be the right UX for some; WVD is a platform option, not a religion.
What we are telling clients this month
- Continue LTSR discipline on existing XenDesktop where stability matters more than novelty.
- For new Azure desktop projects, add WVD preview evaluation as a formal workstream in 2019 planning.
- Accelerate FSLogix learning now — that skill transfers whether you land on WVD, Citrix, or RDS.
- Revisit hardware refresh business cases that assumed on-prem VDI forever; at least run the Azure alternative number.
How this intersects estates we already run
We operate and design for:
- Citrix XenDesktop 7.15 LTSR with NetScaler gateways and classic MCS image cadence
- RDS 2012 R2 + App-V session hosts where economics beat full VDI
- Hyper-V and vSphere capacity that still hosts a lot of those session hosts on-prem
WVD does not delete that installed base. It adds a Microsoft-native Azure path that will attract workload that was waiting for "Windows 10 multi-session without building a Frankenstein." The customers who will move first are those already all-in on Azure networking and identity. The customers who will move later are those whose WAN, profile storage, and broker customizations are deep.
Our multi-tenant Skype and VDI background also makes us cautious about shared control planes. Multi-session density is great until noisy neighbors and noisy profiles collide. Preview testing must include concurrent logon storms, not three engineers clicking around politely.
Competitive dynamics without the fanfic
Citrix and VMware will respond with Azure integrations and messaging about HDX/Blast quality. That is healthy. Enterprises benefit when Microsoft’s platform option forces everyone to justify premium stacks with premium requirements. Our job is not to pick a religion; it is to match requirements:
- Protocol performance on bad networks
- App layering and image factory maturity
- Ops team skill graph
- Licensing already owned
- Timeline and risk tolerance
Sometimes the answer will be WVD. Sometimes Citrix on Azure. Sometimes stay on-prem one more refresh cycle with eyes open.
Risks of announcement-driven projects
We have seen cloud announcements create zombie programs: budget reserved, no success criteria, preview treated as GA, production date set before identity design exists. Avoid that by writing a one-page charter now:
- Decision date after preview evidence
- Workloads in scope / out of scope
- Security and data residency constraints
- Who owns image, identity, and network
- What would make us choose "not yet"
Questions clients ask in the week after Ignite
Should we cancel our Citrix renewal?
Not based on an announcement. Renew with flexibility if you can; evaluate WVD with data in 2019.
Does multi-session Windows 10 mean we can drop Server RDSH?
Eventually maybe for many user types. Today you still need a validated image, profile, and ops model. RDSH remains a known quantity for session density on Server OS.
Is FSLogix required?
For serious non-persistent Windows 10 pools we already treat containers as required practice. Microsoft owning FSLogix makes that easier to justify, not optional to ignore.
Can we put this in the government cloud?
Watch the product availability matrix as preview and GA land. Do not assume commercial Azure features appear everywhere on the same day.
Who runs this — infra or EUC?
Both. Identity and network are shared services; image and pool design are EUC; FinOps is finance. Name the RACI before the pilot so the pilot has somewhere to live.
What we will publish next
When public preview is usable, expect a build notes post: host pool setup, domain join choices, first FSLogix share layout, and the first things that break. Until then, treat keynote demos as direction, not documentation.
If you're facing this
If Ignite left you with a mandate to "move VDI to Azure" and no criteria, slow down. Multi-session Windows 10 in Azure is a real shift; the operational details will decide whether it is your production path or a pilot that dies. When preview opens, we will publish build notes. Until then, we help teams design decision matrices so licensing excitement does not outrun architecture.