What Is VDI and End-User Computing?

Four Words People Use Interchangeably (and Shouldn’t)

Walk into any IT organization and you’ll hear “VDI,” “virtual desktops,” “RDS,” and “DaaS” used as if they’re the same thing. They aren’t, and the differences drive real decisions about cost, compliance, and user experience. This lesson gives you the vocabulary the rest of this course — and every conversation you’ll have as an EUC engineer — depends on.

VDI (Virtual Desktop Infrastructure)

In true VDI, each user gets their own virtual machine, running its own copy of a desktop OS (typically Windows 10 or Windows 11 Enterprise). That VM is isolated at the kernel level from every other user’s VM. This isolation is expensive — you’re paying for a full OS instance, patching, and compute per user — but it buys you application compatibility (anything that runs on a real desktop OS runs here) and a strong compliance boundary, because one user’s process, malware, or crash cannot touch another user’s session.

Session-based computing (RDS / Terminal Services)

Here, many users share a single Windows Server OS instance, each getting an isolated session rather than an isolated machine. This is dramatically cheaper per user — a single server can host dozens of sessions — but application compatibility is narrower (apps must tolerate multi-user execution), and a resource-hungry or crashing process can degrade the experience for everyone sharing that host. Azure Virtual Desktop’s signature move is Windows 11 Enterprise multi-session — the only place you can legally run a multi-user Windows 11 client OS, blending session-based economics with a client OS experience.

Application virtualization

This delivers individual applications, not a whole desktop. The app is packaged, isolated from the underlying OS, and streamed or attached on demand — think MSIX app attach on Azure Virtual Desktop, or RemoteApp publishing. It’s usually layered on top of VDI or RDS rather than competing with them: you might publish only Excel and a line-of-business app to a contractor instead of a full desktop.

DaaS (Desktop-as-a-Service)

DaaS is not a different rendering technology — it’s a consumption and ownership model. In classic on-premises VDI, you own and operate the hypervisor, the broker, the gateway, and the storage. In DaaS (Windows 365, Citrix DaaS, Omnissa Horizon Cloud), a vendor owns and operates most or all of that control plane, and you consume virtual desktops as a managed, often per-user-per-month subscription. Windows 365 pushes this furthest: it is genuinely SaaS, with fixed-price Cloud PCs and no infrastructure for you to size or patch.

A Short History, in Eras

You don’t need trivia dates memorized, but you do need the shape of the story, because it explains why today’s products look the way they do.

1. The multi-user era (mid-1990s). Citrix WinFrame added multi-user session sharing to Windows NT through its MultiWin technology — the first practical way to put many users on one Windows server.
2. The Microsoft licensing era (late 1990s–2000s). Microsoft licensed that multi-user technology and shipped it as Windows Terminal Server, which became Terminal Services and eventually Remote Desktop Services (RDS) — a first-party Microsoft product that Citrix then added value on top of (protocol acceleration, profile management, provisioning).
3. The hypervisor/VDI era (2000s–2010s). VMware, and later Citrix XenDesktop, popularized true VDI: one VM per user, orchestrated by a connection broker, running on-premises hypervisors.
4. The cloud DaaS era (2019–today). Microsoft launched Windows Virtual Desktop (now Azure Virtual Desktop) in 2019, followed by Windows 365 Cloud PCs in 2021. Citrix and the former VMware Horizon business (now Omnissa, spun out independently in 2024) both shipped cloud control planes — Citrix DaaS and Horizon Cloud — while keeping on-premises options alive for regulated and latency-sensitive customers.

The throughline: each era pushed more of the operational burden (hardware, patching, broker infrastructure) off the customer and onto a vendor or cloud platform, while the fundamental problem — deliver a secure, consistent desktop to a device you may not control — never changed.

Why VDI and EUC Matter in 2026

The business case for VDI has shifted from “remote access convenience” to genuine strategic infrastructure. Five drivers show up in almost every real deployment you’ll work on:

• Hybrid and distributed work. Employees expect to work from home, the office, or a client site interchangeably, on whatever device is nearest, without IT re-imaging a laptop for every scenario.
• Security and data-egress control. This is the big one for regulated industries. In VDI/DaaS, data never leaves the datacenter — only encrypted pixels, audio, and input events cross the network. A lost or compromised endpoint device exposes nothing, because nothing was ever stored on it. Financial services, healthcare, defense, and government all lean on this property specifically, not just remote access.
• Contractors and BYOD. Standing up a Cloud PC or a pooled desktop for a short-term contractor takes minutes and is trivially revocable, versus procuring, imaging, shipping, and eventually recovering a physical laptop.
• Regulated industries and audit posture. Centralizing desktops makes patching, logging, session recording, and access reviews consistent and auditable across the whole fleet, rather than trusting thousands of physically dispersed endpoints to stay compliant.
• GPU and AI workstation demand. A newer driver: CAD, media production, and AI/ML development increasingly need GPU-backed compute that’s expensive to put in every employee’s hands as a physical workstation. GPU-enabled VDI (Azure NV-series, NVIDIA vGPU) lets organizations centralize that expensive hardware and allocate it on demand.

As you move through this course, keep coming back to this list. Every platform choice, every architecture decision, and every security control in later modules exists to serve one or more of these five drivers — and a good EUC engineer can always explain which one a given design decision serves.