Virtualisation & VDI Representative engagement

300 desktop sessions without a per-seat licence, and a login time under 20 seconds

VDI — open source / self-hosted

Client
Professional services firm · UK
Region
Two sites, hybrid workers
Scale
300 named users · 180 concurrent sessions · 4 node cluster
Timeline
12 weeks, two pilot groups

Where they started

Staff used laptops with local data, and the firm needed centralised control over information without a per-seat virtual desktop licence that would have doubled the project cost.

The problem, precisely

Desktop performance complaints kill VDI projects. Login storms and profile sprawl had to be solved before rollout, not after.

What we did

Instrumented login time from the first pilot and treated it as the only acceptance metric that matters, then tuned storage and session brokering until it held under a simulated morning rush.

Delivered

  • Golden image pipeline with versioned images and a tested rollback
  • Session brokering HA behind a VIP, surviving a node loss
  • Profile and data redirection onto clustered storage, no local data on endpoints
  • Login-storm test: 180 concurrent sessions started within a defined window
  • Endpoint strategy for the devices staff kept, with documented data rules

Technologies used

  • Proxmox VE + Kasm Workspaces
  • Ceph RBD for session volumes
  • Golden image pipeline + Packer
  • FreeIPA for identity
  • HAProxy session brokering
  • NFS/SMB profile storage

Measured result

Every number below was measured and reported to the customer. Where we do not have a measured figure, the entry is a fact rather than a percentage — we would rather be precise than impressive.

  • 17s p95 login time under simulated login storm
  • 180 concurrent sessions supported on 4 nodes
  • 0 per-seat virtual desktop licences purchased
  • 100% of user data relocated to controlled storage
The reusable part

Design VDI against the login storm, not the average day. Almost every VDI failure is storage saturation in the first fifteen minutes of the morning.

Something here look familiar?

Most of these started with a cluster audit: two weeks, fixed price, and a risk map that names the single points of failure before you commit to any work. Keep the report either way.