This article is based on a case study originally published on citrix.com, describing a real-world deployment completed in the first half of 2019 by Citrix and SNP Technologies.
Background: The Challenge for Graphics-Intensive Industries
Architecture, design, and engineering firms depend on demanding applications such as Autodesk and SolidWorks. These tools require significantly more GPU and CPU resources than typical business software, which creates two persistent pain points for IT teams serving this sector:
- High-end workstation endpoints with dedicated graphics cards carry a steep capital cost.
- Decentralized endpoints make it difficult to centrally manage data security and compliance.
Connecticut-based Citrix Service Provider (CSP) SNP Technologies approached the Citrix CSP team with a goal: build a fully cloud-hosted offering targeting this niche vertical, eliminating expensive local hardware while keeping sensitive design data centrally secured.
Solution Architecture: Citrix Cloud + Azure N-Series VMs
As a 100% cloud-focused partner, SNP Technologies built their offering entirely on Citrix Cloud Virtual Apps and Desktops service. This eliminated the need to deploy and maintain on-premises Citrix infrastructure, reduced operational overhead, and made onboarding new customers significantly faster.
For compute, the solution used GPU-enabled Azure N-Series virtual machines running Windows 10. Each end user receives a dedicated, high-end Windows 10 virtual desktop pre-installed with the engineering applications they need. All data remains locked down in the cloud, managed from a single location.
Controlling Compute Costs with Citrix Autoscale
A natural concern with GPU-enabled VMs is cost. Azure N-Series instances are among the more expensive VM SKUs. Citrix addressed this directly through its Autoscale feature, which is built into the Citrix Cloud Virtual Apps and Desktops service at no additional charge.
Autoscale provides proactive power management of virtual machines based on:
- Load — machines scale up or down based on active session demand.
- Schedule — machines are powered off during off-hours and weekends when users are not active.
- A combination of both — offering the most cost-efficient coverage for varying usage patterns.
Autoscale supports Microsoft Azure Resource Manager as a VDA hosting platform, making it a natural fit for this deployment. Because there are no on-premises components, SNP Technologies can manage the entire environment from the Citrix Cloud console.
Optimizing the User Experience: HDX Graphics Policies
Delivering a smooth, responsive experience for graphics-intensive applications over a virtual session requires careful tuning of Citrix HDX policies. The Citrix CSP team arranged a working session between SNP Technologies and a Citrix HDX graphics product manager. After multiple testing cycles, the following policy configuration was found to be optimal for this deployment type:
-
Optimize for 3D Graphics Workload — Enabled
Configures default settings suited to graphically intensive workloads. Should only be applied when a GPU is available to the session. -
Hardware Encoding for Video — Enabled
Allows the GPU to compress screen content using the H.264 video codec, offloading encoding from the CPU. -
Video Codec for Compression — For the Entire Screen
Applies H.264 compression to the full screen rather than selectively. Recommended when the session involves heavy server-rendered video and 3D graphics, as it improves both user experience and bandwidth efficiency. -
Target Frame Rate — 60 FPS
Sets the maximum frames per second transmitted from the virtual desktop to the client device. Increasing this from the default of 30 FPS improves visual smoothness for fast-moving 3D content, at the cost of additional bandwidth. -
Visual Quality — High
Controls the image quality level delivered to the endpoint. The “High” setting is recommended when visually lossless quality is important, as it was in this architecture and engineering context. -
Graphics Status Indicator — Enabled
Displays a status overlay in the user session showing active graphics mode details, including provider, encoder, hardware encoding status, image quality level, progressive display status, and lossless text. Useful for both troubleshooting and validating that GPU acceleration is active.
Important: These settings represent what worked best for this specific use case. The optimal policy combination will vary depending on the applications in use, available bandwidth, and client device capabilities. Work with a Citrix engineer to tune policies for your specific environment.
Secure Remote Access with Citrix Gateway Service
SNP Technologies also leveraged the Citrix Gateway service, which is included with Citrix Virtual Apps and Desktops licenses. Citrix Gateway provides a globally distributed network of points of presence (PoPs), routing users to the nearest available node to minimize latency.
By using the Gateway service rather than deploying self-managed NetScaler appliances, SNP Technologies was able to include enterprise features — including high availability and global server load balancing — in their branded offering without standing up or maintaining complex networking infrastructure.
Outcomes and Key Takeaways
The result of this collaboration was a fully cloud-native, GPU-accelerated virtual desktop offering that SNP Technologies brought to market successfully. Key outcomes included:
- Faster customer onboarding — new customers can be provisioned in hours rather than days or weeks.
- Eliminated endpoint hardware costs — end users connect from thin clients or standard devices without needing local GPU hardware.
- Centralized data security — all design data stays in the cloud, reducing the risk of data loss or theft from endpoint devices.
- Controlled compute spend — Autoscale keeps Azure N-Series VM costs in check by powering down idle machines automatically.
- No on-premises infrastructure — the entire solution runs on Citrix Cloud and Azure, consistent with SNP Technologies’ cloud-first mandate.
For Citrix Service Providers evaluating how to address graphics-intensive verticals — including architecture, engineering, media production, or geospatial analysis — this case study demonstrates that a fully cloud-hosted GPU delivery model is both technically viable and commercially feasible when the right cost controls are in place.
Source: Citrix.com. Original content describes a deployment completed in partnership between Citrix and SNP Technologies during the first half of 2019.