EUCCOST05-BP04 Choose an appropriate running mode for your EUC workload where applicable - End User Computing (EUC) Lens
Implementation guidance

EUCCOST05-BP04 Choose an appropriate running mode for your EUC workload where applicable

Amazon WorkSpaces can be used with monthly and hourly pricing, while Amazon WorkSpaces Applications supports Always-On, On-Demand, and Elastic fleets. Choosing an appropriate running mode can significantly impact the cost of your EUC services. Historical usage data (usage patterns) of a reference environment can help you assess which running mode to use for your EUC workloads. 

Level of risk exposed if this best practice is not established: Medium

Implementation guidance

When you use Amazon WorkSpaces, you can choose between Always-On and On-Demand running modes, which translate into monthly and hourly billing respectively. For the non-GPU bundles, there is a breakeven point at roughly 80 hours of usage per month, at which point the Always-On WorkSpace will be more cost-effective. If your users use their WorkSpace for less than 80 hours per month, the On-Demand running mode is usually the more cost-effective model for non-GPU bundles.

You can deploy the Cost Optimizer for Amazon WorkSpaces to get reports with recommendations on which running mode to select for your WorkSpaces and automatically convert your WorkSpaces to the most cost-effective running mode. For the GPU bundles, the breakeven point varies from bundle to bundle. The Amazon WorkSpaces Pricing page helps you calculate the breakeven point for these bundles.

Amazon WorkSpaces Applications offers three different fleet types: Always-On, On-Demand, and Elastic. Explore the fleet types to determine the right balance between cost-effective operation and desired user experience.

  • With Always-On fleets, your fleet instances will constantly be running while the fleet is in a started state, and you'll be charged the respective instance fee per hour per instance in your fleet. 

  • On-Demand fleets have those fleet instances not in use in a stopped state, for which you'll be charged the lower stopped instance fee per hour per stopped instance in your fleet.

    • This can make a significant difference to your cost, especially when your fleet instances are higher-end instances.

    • However, using On-Demand fleets will prolong the logon time by up to 120 seconds.

    • Both Always-On and On-Demand fleet instances are charged on one-hour increments, while Elastic Fleet instances are charged on one second increments, with a minimum of 15 minutes.

  • As opposed to Always-On and On-Demand, Elastic fleets do not require you to manage scaling policies and provision buffer capacity, since the pool of Instances in an Elastic fleet is managed by WorkSpaces Applications.

Amazon WorkSpaces Applications offers multi-session fleets, which allow multiple users to use a single WorkSpaces Applications fleet instance. Depending on the user density you can achieve on a given instance, you may be able to further optimize your WorkSpaces Applications costs compared to a single-session fleet. If you plan to use multi-session fleets, consider resource requirements, instance specifications, and user behavior. For specific guidance, see Multi-Session Recommendations .