unable to load Concurrent users (8-10 users) in Open source R studio of R version 4.1.0 within 6 seconds

[arunatindium]

I want to have concurrent users about 8 - 10 users with decent load speed (opening application in 5 - 6 secs) for open source RStduio server of R version 4.1.0
Currently , the app opens in 15 secs for just one user but when we start adding other users the app starts to slow down (takes a minute or so even for 4 users) . I understand that R studio open source allows only one session and in that we can use concurrent users.

Is it possible to have concurrent users for about 8-10 users with decent speed (5-6 secs).
We have the Rstudio in EC2 instance of t2.xlarge . No docker used. No load balancer , we have just one EC2 instance of t2.xlarge

I did see a post for similar issue at Multiple rsessions for a single user - #3 by mksf but this is an old post and it closed too.

Please provide your input in this regard.

[pieterjanvc]

Hi,

Welcome to the RStudio community!

I'm not an expert on the server parts, but if you like to have a Shiny app run for multiple users, it's better to use ShinyServer. In Rstudio, you can access the same running Shiny app multiple times, but it will all be run on a singe processor. If you have an app that requires a lot of CPU or memory, this will be inefficient. Shiny server can use multiple CPU's to run (I'm not sure if the free edition can).

Another important thing to consider is optimising your app to make sure there are no redundant calculations or memory allocations when multiple users are working. Look at the scoping article for more info.

Good luck
PJ

[andresrcs]

@pieterjanvc

Shiny server can use multiple CPU's to run (I'm not sure if the free edition can).

The open source version is limited to a single R session (thus only one core is used) , you can though make use of other open source deployment options to overcome this limitation like ShinyProxy

But I think the OP is talking about RStudio Server sessions and not shiny apps, I think the only way around the open source version limitations is to use containers and an orchestrator like Kubernetes for example.