ArcGIS Pro uses modern computing hardware to render graphics-rich maps and 3D scenes, and to perform complex spatial analysis. When the right resources are provisioned, ArcGIS Pro can perform as well in a virtualized or cloud environment as on a powerful desktop workstation.
Virtualization allows organizations to provide secure, scalable access to ArcGIS Pro across different user groups. Whether on-premises or in the cloud, a successful deployment depends on ensuring that the underlying virtual machines (VMs) are configured to meet the ArcGIS Pro system requirements.
To achieve this, the virtualized environment must provide the resources that ArcGIS Pro requires, based on its architecture. The following are the primary components of that architecture, along with descriptions of how they exist in virtualized environments:
- Central processing unit (CPU)—ArcGIS Pro is a multithreaded application that uses as many CPU cores as are available. Tasks such as geoprocessing, data input and output, and map rendering benefit from higher core counts and clock speeds. In a virtual environment, virtual CPUs (vCPUs) are presented to the VM and used in the same way as physical cores.
Tip:
If performance is slow when running multiple tasks, ensure that your VM template includes enough vCPUs. Higher clock speed (GHz) per core can also help performance.
- Random Access Memory (RAM)—An adequate amount of RAM is important for performance on both physical and virtual machines. This is necessary because in a virtual environment, vCPUs are presented to the VM and used in the same way as physical cores. ArcGIS Pro loads datasets, layers, and cache into memory while running processes. Allocate an appropriate amount of RAM based on the ArcGIS Pro minimum and recommended requirements, and include additional capacity if your workflows involve large datasets or 3D scenes.
Tip:
If map drawing is slow, or your VM pauses during analysis processes, check to see whether your session is running low on memory.
- Graphics processing unit (GPU)—ArcGIS Pro uses the DirectX or OpenGL graphics application programming interfaces (APIs) to deliver an optimal graphics experience in which panning in maps and scenes is smooth and fluid. In virtual environments, the GPU resources can be delivered in one of the following ways:
- Shared GPU (vGPU)—The GPU is virtualized and shared between multiple VMs. This implementation provides a balance of scalability and performance for most users. This type of GPU is recommended for mapping and analysis work.
- GPU Passthrough (dedicated GPU)—A full, physical GPU is assigned directly to a VM. This approach offers the best performance and is recommended for users working with 3D visualization, animation, or very large datasets. However, this approach is less scalable because one GPU can only serve one VM.
Tip:
If 3D navigation is not performing as expected, check whether the VM is provisioned with the correct vGPU profile or whether a passthrough is needed.
These hardware resources must be configured independently in each virtualized environment. For on-premises virtualized environments, configuration is managed and maintained by virtualized environment administrators. For cloud-based environments, administrative responsibilities are managed by the cloud provider. In cloud-based environments, you are responsible for selecting appropriately sized and configured instances. The on-premises virtualization and cloud virtualization sections of the ArcGIS Pro help provide requirements for each environment and describe how the configuration is managed.
Remote Desktop Session Host solutions
A Remote Desktop Session Host (RDSH) or Citrix Virtual Apps environment supporting multiple user connections to multiple sessions of ArcGIS Pro is not supported. Both RDSH and Citrix Virtual Apps deliver applications from a shared Microsoft Windows Server operating system where multiple user sessions compete for the same hardware resources.
While it is technically possible to deliver 3D applications on a Windows Server operating system (OS), RDSH is not designed for high-performance 3D workloads. The way the OS manages GPU memory and scheduling limits both scalability and consistency, resulting in degraded performance when multiple users share the same GPU.
The following are some reasons that RDSH and Citrix Virtual Apps are not supported environments:
- Intended use—RDSH environments are optimized for 2D applications such as ArcMap, which have been successfully deployed in shared server environments for years.
- 3D application limitations—Running 3D applications, such as ArcGIS Pro in 3D mode, is possible but not ideal. User density and performance consistency degrade because the Windows Server OS does not efficiently manage GPU memory or scheduling among concurrent sessions.
- GPU management constraints—There are no vGPUs at the Citrix Hypervisor level to control or schedule the GPU time across users. As a result, GPU memory and compute resources cannot be isolated or balanced effectively between sessions.
- GPU resource behavior—The GPU is passed through directly or to the OS, and each user session consumes GPU resources freely until GPU memory or processing capacity is fully utilized. When this occurs, new sessions may struggle to start or may not perform as expected.
- Performance degradation under load—As additional users connect, they may need to wait for GPU-intensive processes to complete, or for the system to offload GPU tasks to the CPU. This offloading leads to noticeable slowdowns and degraded performance for all users on the server.
- Scalability implications—Because GPU resources cannot be efficiently shared or scheduled, user density and overall performance stability for 3D workloads in RDSH are significantly lower than in environments with dedicated or vGPU-enabled single-session desktops.
In short, although RDSH can technically host 3D applications, it is not designed for sustained or scalable 3D workloads. The lack of Citrix Hypervisor-level GPU management and resource isolation leads to contention, performance degradation, and reduced user capacity. For organizations delivering GPU-intensive 3D applications such as ArcGIS Pro, dedicated GPU-backed VMs or a vGPU-enabled environment are strongly recommended to ensure consistent performance and a better user experience.