The difference between “Always-On” and “On-Demand”
Choose between Always-On and On-Demand fleets to best meet the needs of your students, faculty, staff, and your institutional priorities. Always-On fleets provide immediate access to resources and require a number of machines to be in a running state even if no users are connected. This is recommended when your users need high availability and instant access to applications. On-Demand fleets begin when a user connects to an instance within the fleet, similar to your laptop coming out of hibernation. This is best when your users can wait up to 2 minutes to start applications and for streaming applications that have sporadic use.
| On-Demand | Always-On | |
|---|---|---|
| Instances | Stopped | Running |
| User session start | Up to 2 minutes | Instant on |
| Optimized for | Up to 2 minutes | Instant availability of apps |
| Use cases | Cost savings are critical | Need for instant availability |
These configuration choices directly impact Amazon Web Services (AWS) billing. Always-On provides instant access and in turn requires paying full hourly usage prices, based on the number of instances in the pool whether or not the instance is used. On-Demand provides a slightly delayed access while paying a significantly smaller fee for the standby instance; the full hourly usage fee is only incurred when a user logs in. By default, Propeller utilizes the On-Demand configuration to support your cost-savings.
Sample AWS Pricing*
| On-Demand | Always-On | |
|---|---|---|
| Number of instances | 5 | 5 |
| Instance configuration | 4vCPU / 30.5 GB of Memory | 4vCPU / 30.5 GB of Memory |
| Cost per idle hour | .02 | -- |
| Cost per running hour | .50 | .50 |
| Usage hour per instance per month | 10 | 720 |
| Total cost per month | $91.20 | $1,800 |
* AWS list pricing for AppStream 2.0, all costs are approximate.
Comments
0 comments
Article is closed for comments.