Australian cloud for business software — built and run in Sydney since 2002

NetLeverage
NetLeverage

Shared office PCs reaching hosted applications safely

Technical article
Hosting

About this article: This article is reviewed against the cited sources. This is source research, not lab validation. Check the stated versions and your own requirements before making changes.

A shared office PC—reception, ward station, dispatch bench—should let each staff member reach hosted applications without leaving the previous person’s documents, browser sessions, or cached passwords on screen. Safe patterns combine per-person sign-in, session isolation on the remote farm, and clear rules about what may stay on the local device.

Why it matters

Clinics and warehouses optimise floor space with shared desks. If user B inherits user A’s Windows profile or an unlocked hosted session, you inherit privacy incidents and data-integrity risk. This article addresses the operating model, not Client Access Licence procurement (a different decision track).

How it works

Hosted applications run in remote sessions on RD Session Host servers. Microsoft notes that multiple users typically share a session host and “must sign in to collections with non-administrative accounts.” Each remote session is tied to the user’s credentials on the farm; ending or signing out of the session reduces cross-user bleed on the hosted side.

The local shared PC is still a risk surface: auto-saved files on C:, browser passwords, and USB devices. Microsoft’s gateway documentation discusses optional client computer group requirements in Connection Authorization Policies—meaning some designs can bind access to known device groups. Whether that fits a generic shared PC depends on domain join and operational discipline; it is not a universal default.

Windows Shared PC mode helps organisations configure a separate local multi-user PC for shared desks; it is not the same product as RDS. When you use Shared PC mode, follow Microsoft’s guidance for configuring shared or guest Windows devices locally, and still decide named vs shared Windows accounts for accountability.

Publishing RemoteApp alone is not a security boundary. A single application tile still runs in a full session context; pair narrow publish scope with least-privilege gateway policies, application permissions, and redirection controls appropriate to the kiosk.

Labelled example — medical reception: Each staff member signs into Windows with their own account, launches hosted practice software, signs out of Windows at shift end; no shared “reception” password.

Labelled example — warehouse clock PC: A locked-down local kiosk profile may launch only the published RemoteApp tile, but each operator must sign in to the hosted farm with a personal identity (not a shared application login) so sessions are attributable; clipboard and drive redirection disabled by policy where required.

What to choose and why

Model Trade-off
Named local Windows accounts Best audit trail on device; more login steps
Shared local account + personal remote sign-in Faster locally; relies on remote sign-out discipline and personal hosted identities
Published RemoteApp only (no full desktop) Smaller task surface if combined with policy—but not a boundary by itself

Decide named vs shared local Windows accounts first for the kiosk PC. Hosted farm sign-in must still be a personal identity for every operator—do not use a generic shared hosted account. Train shift handover: sign out of the hosted session, close the client, lock or sign out of Windows.

Risks and limits

Local profile caching, browser “remember me”, and redirected drives can leak data even when remote sessions are clean. No operational test was run on a clinic-style device.

Sources

Related reading

Learning notes

  1. Remote session isolation helps on the server; local PC hygiene is still mandatory.
  2. Shared desks need explicit sign-out culture, not only policy PDFs.
  3. RemoteApp narrows tasks but still needs least-privilege and app controls—not a security boundary alone.
  4. Gateway CAPs can reference device groups—design, not assumption.
  5. CAL mode choice is related but not answered here.