Shiny Server Pro has a limit to the number of concurrent users it can serve. A concurrent user is defined as a single individual with an active browser connection to one or more customer applications (see 3.2.1 of the End User License Agreement).
More plainly, a concurrent user can be described as one human on one browser connecting to one instance of Shiny Server Pro.
The standard license of Shiny Server Pro comes with a limit of 20 concurrent users. When the number of concurrent users exceeds 20, additional users will see an error message stating that the license quota has been reached (see below). Note, this error message can be customized by modifying custom templates as described in the admin guide.
A single concurrent user can have multiple connections to Shiny Server Pro. Every browser tab that requests a Shiny app makes a connection to Shiny Server Pro. When a user runs one or more apps on multiple browser tabs, that user creates multiple connections to Shiny Server Pro.
Example 1: A user connects to a single Shiny app. Outcome: One concurrent user and one connection.
Example 2: A user connects to the same Shiny app two times by opening two tabs in a browser. Outcome: One concurrent user and two connections.
Example 3: A user connects to two Shiny apps via two tabs in a browser. Outcome: One concurrent user and two connections.
Example 4: A user connects to two Shiny apps using two different browsers. Outcome: Two concurrent users and two connections.
Example 5: A user connects to two Shiny apps on a load balanced cluster and one app is served from one instance and another app is served from another instance. Outcome: One concurrent user and one connection on each instance for a total of two concurrent users and two connections.
You can increase the limit for concurrent users on a single instance of Shiny Server Pro by purchasing additional user packs. The standard pack comes with an additional 20 concurrent users. The large user pack comes with an additional 150 concurrent users.
For information about how to scale and tune applications see this article.
Comments