I would like to know if it is possible to run an RPA inside a Windows Docker container.
Currently RPAs are limited to running one instance per machine. It would be much more cost effective if we could spin up multiple docker containers that are bundled with their own version of the Pega runtime and have multiple RPAs running on a single machine. Does anyone have experience doing this?
I am not familiar with docker container, however a quick search mentioned that it runs the server kernel. Currently, there is no explicit support for that version of Windows (Server 2016). In addition, running user applications on an OS that differs from where they are currently run is problematic since the applications and websites they use are likely not tested on that OS (so even if PR ran fine on the OS, the applications that you wish to automate may not, or at the very least may work differently). A better method to achieve a similar result would be to use virtual machines running the same OS as the users whose tasks are being automated.
I believe many of the applications we have would run fine in the docker container on windows server 2016. I think if Pega worked on adding support for Windows server 2016 (especially for RPAs) then a lot of your customers would take advantage of running them in a docker container. VMs take longer to set up, don't use resources as efficiently, and in my company's case, cost us more money. Docker containers are also a lot easier to deploy (it would be nice to not have to install the runtime on every VM we run an RPA on and instead just put our RPA package in a container that already has the runtime) plus they can be stored in an artifact repository just as easily as a .openspan file. I hope this is something yall would consider putting on your radar. If it isn't possible then we will survive but it would make testing, deploying, and managing RPAs a lot easier.
Came across this past discussion regarding windows container/Docker. We have plan for doing POC to package Pega Robotics runtime with other applications as a container. Have you tried it earlier and worked?