Options for Deploying Desktops in the Cloud

2010
3
07-02-2019 01:11 PM
Labels (1)
Esri Contributor
7 3 2,010

As cloud adoption evolves from Web GIS to full GIS deployments, questions continue to be raised such as, “What about the desktops?”. That is, when moving desktops to the cloud, what technologies should be used to support Esri desktop GIS? The cloud offers multiple desktop options and the following will provide some high-level guidance as to how and when these technologies should be used. It is important to realize that each deployment is unique and deciding on which of these technologies to deploy involves multiple factors. The purpose of providing this information is primarily to share information regarding all of the potential solutions and high-lighting some of their key characteristics. However, they will not be ranked in any way as deciding on one approach over the other requires more detailed analysis and discussion based on specified requirements, costs, and constraints. Further, this list can likely be expanded but the solutions below represent the most common options that Esri has encountered. 

Note: ArcGIS Pro requires a GPU-enabled machine type for the underlying host VM. Examples include the NV6 for Azure and a Graphics Design Instance for AWS AppStream.

Virtual Machines - Azure and AWS

  • Use Case: Typically used to support administrative functions or small number of desktop users
  • Client Connectivity: Utilizes the Remote Desktop Connection client and the RDP protocol
  • User Experience: Published desktop with growing visual latency as geographic distance increases
  • Scalability: Limited due to no more than two concurrent users per VM
  • Management: Typically deployed without a base image
  • User Profiles: Locally stored per VM

Remote Desktop Services - Azure and AWS

  • Use Case: Supporting users at scale where the users are not globally distributed
  • Connectivity: Utilizes the Remote Desktop Connection client and the RDP protocol
  • User Experience: Published desktop or apps with growing visual latency as geographic distance increases
  • Scalability: Limited for ArcGIS Pro based on the number of concurrent sessions that can share a GPU
  • Management: Can be used with snapshot technology to create a base image
  • User Profiles: Roaming profile or equivalent, assuming at least two servers deployed

Citrix Virtual Apps and Desktops (XenApp) - Azure and AWS

  • Use Case: Supporting users at scale where the users could be globally distributed
  • Connectivity: Utilizes the Citrix Workspace app and the HDX protocol
  • User Experience: Supports both published desktops and apps and performs well with high-latency
  • Scalability: Limited for ArcGIS Pro based on the number of concurrent sessions that can share a GPU
  • Management: Can be used with snapshot technology to create a base image
  • User Profiles: Roaming profile or equivalent, assuming at least two servers deployed
  • Other: Can utilize Citrix Cloud to manage the "back-end" (e.g., Controllers/Licensing)

Amazon WorkSpaces - AWS

  • Use Case: Supporting users at scale where client bandwidth is not a limiting factor
  • Connectivity: Utilizes either a desktop or web client with the PCoIP protocol
  • User Experience: Supports a published desktop to an assigned WorkSpace instance
  • Scalability: Can scale as needed as users increase but is 1:1 user to VM assignment
  • Management: Cannot be used with snapshot technology so each WorkSpace is an independent deployment
  • User Profiles: Locally stored on each WorkSpace

Amazon AppStream 2.0 - AWS

  • Use Case: Supporting users at scale, but not all use cases are known since this is a newer offering
  • Connectivity: Utilizes either a desktop or web client with the NICE DCV protocol
  • User Experience: Supports published desktop applications
  • Scalability: Can scale as needed as back-end infrastructure capacity is managed by AWS
  • Management: Based on creating base images for different application configurations as needed
  • User Profiles: Saved to a Virtual Hard Disk (VHD) and synchronized to Amazon S3
  • Other: Esri / AWS AppStream 2.0 Deployment Guide

Windows Virtual Desktop - Azure*

  • Use Case: Supporting users at scale, but not all use cases are known since this is a newer offering
  • Connectivity: Utilizes the Remote Desktop Connection client and the RDP protocol
  • User Experience: Supports both published desktops and apps
  • Scalability: Can scale as needed and either be deployed as one user per VM or concurrent sessions
  • Management: Can be used with snapshot technology to create a base image
  • User Profiles: Roaming profile or equivalent, assuming at least two servers deployed
  • Other: The only solution supporting multi-session Windows 10

* Third-party vendors are working to extend the core Windows Virtual Desktop capabilities and providing additional client options and management features. Current examples include CloudJumper Cloud Workspace and Citrix Managed Desktops.

3 Comments
Occasional Contributor II

We use AWS to run our desktop products and connect to our enterprise server at another site.  Keep in mind that ArcGIS Pro requires the graphics package from AWS and the price is quite a bit more.  I'm currently running the non-graphics package with 16GB of memory and it runs all of the desktop products very well, but ArcGIS Pro struggles a bit.  It's usable for doing table edits and running some geoprocessing tools, but it bogs down quite a bit when viewing the map.  I really wish ESRI would separate out the the 3D fuctionality of Pro so those of us that don't need it can run it on machines w/o a graphics card.

Esri Contributor

Thanks David. I will update the post regarding Pro requiring a GPU. Yes, you can run Pro without a GPU but what you have probably noticed is that your CPU utilization has increased dramatically. I have seen it increase 10x so yes, its possible for a single user but not a scalable approach.

New Contributor III

Agreed!

About the Author
I joined Esri in 1998 and I am currently a senior Enterprise Solutions Architect within Esri's Implementation Services Department. I have over 30 years of experience leading teams, supporting sales initiatives, providing technical management, architecting systems, implementing systems, evaluating technology, and providing technical consulting to meet customer business objectives.