What is too big? Too many concurrent jobs? We found the integration with AWS to autolaunch workers to be quite easy to setup.