r-knative
Example of building R APIs for Knative
Using something like R-Plumber it is very easy to create APIs for R-scripts and run statistical computing and graphics on a Kubernetes cluster using Knative.
Looking at api.R you can see examples of how you can define endpoints for your applications and still leverage all the power and ecosystem or the R Language inside containers with minimal effort. As an example:
#* Plot a histogram
#* @png
#* @get /plot
plot <- function(){
rand <- rnorm(100)
hist(rand)
}
Now you can build a container and deploy this endpoint that will auto-scale and receive events (run serverless) with 3 basic steps:
docker build . t <your/containername># build your local script into the containerdocker push <your/containername># push your container image to a registrykn service create my-r-api --image=<your/containername> --limits-cpu=0.5 --limits-memory=500M# create a Knative Service
This will give you an API endpoint where you can access the plot function to access your API.
The R-Plumber website has more details about more complex applications that can be built using the framework.