Top Design Decision Guidance: Deliver Citrix DaaS in Azure on AMD Compute
August 17, 2023
Recently, the teams at Microsoft, Citrix, and AMD collaborated to provide guidance in making informed decisions about selecting the right instance types for Microsoft Azure hosting Citrix Workloads.
The collaborative research focuses on optimizing instance types for hosting Citrix Desktop as a Service (DaaS) workloads and offers insight on choosing appropriate AMD compute instances, both with and without GPUs, in the Azure environment.
Here is an overview of the platforms involved in this research:
- Azure Platform: Microsoft’s Azure provides a cloud computing platform where physical hardware resources, such as computers, storage drives, and networking, are virtualized and made accessible through the internet or direct connections to Microsoft’s data centers worldwide. These data centers are organized into regions, which offer redundancy in case of failures and reduced latency by locating resources closer to users.
- Citrix DaaS: Citrix’s Desktop as a Service (DaaS) offering securely delivers various applications and desktops, including Windows, Linux, Web, and Software-as-a-Service (SaaS) applications, to multiple devices. This technology empowers modern digital workspaces by offering advanced management, scalability, and rich multimedia experiences over various networks. The ability to provision Citrix workloads on Microsoft Azure eliminates the need for internal infrastructure, relying instead on Azure’s computing, networking, and storage resources.
- AMD: Advanced Micro Devices (AMD), a leader in high-performance computing, provides technologies that accelerate a wide range of data center workloads. AMD’s EPYC™ processors are known for their energy efficiency and high performance, making them suitable for various computing tasks. The paper explores different AMD instance types available within Azure, such as the NVv4-series instances equipped with AMD’s Radeon™ Instinct™ MI25 GPU, optimized for Virtual Desktop Infrastructure (VDI) and remote visualization scenarios.
The NVv4-series instances come in 4 different sizes:
Determining the Most Efficient Instance Types to Host Citrix DaaS with Login Enterprise
To evaluate the performance and scalability of different instance types, the group employed Login Enterprise. Login VSI’s solution, Login Enterprise, proactively monitors the performance of virtual desktops and applications. By simulating artificial workloads during test runs, the solution generated data to analyze the scalability of various AMD-powered Azure instances.
Login Enterprise helps organizations proactively manage their virtual desktops and applications’ performance, cost, and capacity. The platform is 100% agentless and can be used in all major VDI and DaaS environments, including Citrix and Microsoft. With Login VSI, IT teams can plan and maintain digital workspace ecosystems with less cost, fewer disruptions, and lower risk.
The Login Enterprise offers metrics to determine optimal user capacity for the instance:
End-user experience score (EUX): The EUX Score gauges the quality of user experience during their virtual session by utilizing various metrics and performance measures recorded throughout the session. EUX scores span from 1.0 (indicating the poorest experience) to 10.0 (representing the best experience). Typically, scores falling below 5.5 suggest an unacceptable user experience.
Maximum Recommended Users (VSImax): The calculation of VSImax involves analyzing individual session metrics to ascertain the maximum number of concurrent users a given system can support.
The Study’s Results and Recommendations
The study examines the scalability of different AMD instance types with two types of workloads: the Knowledge Worker and GPU Intensive.
Here are the key findings:
- Windows 10 multi-session instances accommodate 30% fewer users than Windows Server instances.
- While the D4as_v5 instance is the most cost-efficient for Knowledge Worker tasks, D8as_v5 is recommended for a better user experience at a slightly higher cost.
- For GPU-intensive workloads, the NV_v4 instance types offer improved user experiences with GPU support.
The research paper also assessed the end-user experience by considering factors like the End User Experience (EUX) score, noting that the most cost-efficient instance types tend to have slightly lower EUX scores. At the same time, slightly higher costs can result in significantly improved user experiences.
Design Decisions for Delivering Citrix DaaS in Azure on AMD Compute
Making the most optimal design decision requires performance validation for your organization’s specific business requirements. From this study, we can assume that, in general, instance types with lower vCPU counts provide a balanced approach between cost and performance. Considering this study’s insights and using Login Enterprise to validate your design decisions, you can select the most appropriate AMD-powered Azure instance types for your unique workloads.
Ready to make swifter, more strategic design decisions? Contact a Login Enterprise expert today to get started!