In harell/microservices: Expose the Functionality of a Software Program to the Outside via 'HTTP' Requests

source(file.path(usethis::proj_get(), "vignettes",  "_common.R"))
file_fs <- system.file("inst", "configurations", "fs.yml", package = "microservices", mustWork = TRUE)
info <- config::get("use_plumber_microservice", file=file_fs)
microservice <- list(files = info$files, dependencies = as.data.frame(info$dependencies))

How It Works

Given a path to a folder

When use_microservice(path = ".") is called

Then the function creates the following files:

And updates the following files:

And adds the following packages to the DESCRIPTION file:

kable(microservice$dependencies)

When to Use plumber

• A Single user/machine applications.
• Scheduled tasks. For example, you could use AirFlow with HTTP Operators to automate processes.

plumber Advantages

• Comes with familiar way to document the microservice endpoint.
• Maturing package that comes with documentation, examples and support.

plumber Disadvantages

• Runs on a single thread. That means that parallel algorithms such as random forest, can only be run on one core.
• Serves only one caller at a time.
• Can't make inward calls for other services. At least with the current specification (v.1.0.0), plumber can't be re-entrant. For example, if a microservice has three endpoints,read_table, write_table, and orchestrator, where the orchestrator reads a data table, transforms it, and writes it back, then the orchestrator can't make inwards calls via HTTP to read_table and write_table.

Workflow

harell/microservices documentation built on March 2, 2021, 3:15 a.m.