Basic UiPath infrastructure using minimum IT resources

Hello, we try to kick start RPA project using minimum IT resources. I have completed RPA Infrastructure training, also have read almost all related docs. I have a vision but need to clarify it.

2 developers, 1 robot environment:

Install Studio in developer PC (Windows 10 / enterprise apps).
Install Studio in test virtual server (Windows 10 / enterprise apps).
Install Orchestrator in the virtual server (Windows Server 2016 / additional apps required).
Install Microsoft SQL Express in the virtual server (Windows Server 2016).
Install Robot in prod virtual server (Windows 10 / enterprise apps).

As far as I know, ElasticSearch + Kibana can be added later as it is not crucial for small ecosystem.
Redis also is for large deployments and cluster installations.

Will appreciate any tips / corrections. Thank You in advance.

Seems good for most part.

from my perspective as a developer…
Here are a few other thoughts I have:

You actually have 2 Robot environments, cause the test would and should be used for Robots. Otherwise, you wouldn’t have an environment for processes to automate until they are ready for production migration. This may depend on your license negotiation, though, and maybe you were planning on just using Developer robots.

It is good practice to have Disaster/Recovery. I think the suggestion is to use a Load Balancer with 2 Orchestrator servers in different data center locations. So when one goes down, the whole system doesn’t go down.

Related, if your Robot server goes down (and it will), you have no way to run jobs on other servers. But, maybe you can use the Test environment.

You will find that the Robot Service that controls whether robots are connected can crash or fail to load on boot up. So make sure the service is set up to automatically start and retry if it stops.

You didn’t specify how many Robots were going to be installed, but as you scale up, each user account will require specific user profile settings. IT never handles this very well. You may consider more of a VDI, where you can have a desktop image that is centralized for each user profile that logs in, so you only need to change the settings for one image.

I am just a forum contributor, but these are just some things that came to mind.


Thank You for prompt reply and great tips. Disaster recovery option we will implement only after this basic infrastructure will work. The reason why I think to set up Studio in test environment - it will be easier to test robotic solution by manually running processes. Because test and prod environment will be as same as possible I presume that if process works in test, we can deploy it to prod with confidence. But I understand that during testing phase it is also important to check how orchestrator will behave with new process (we will not forget this in near future).

1 Like