How many node red instance on 1 server

I am currently looking into big PLC controlling project, witch might could be run with Node Red:
The Case is:

Around 200 pcs of small PLC in different locations should be running with same number of individual dashboards. ( All running in 1 big network ) 1 PLC, for 1 dashboard.
Would it be possible to run all those instance from one big server, or is the best solution to add 200 pcs of Pi's to run NodeRed instance locally next to PLC's ?

There is no real best however there are things you can do to improve manageability.

1stly, I doubt whether 200 would be sensible on 1 server. Nor would 200 individual devices to maintain and update. Then there is keeping your device flows up to date (managing backups etc etc, the list goes on).

I would personally recommend you look at what others have done to smooth this process out. For example, the company I work for (FlowForge) has an open source product that can take away most of these headaches (like spinning up instances of node-red, seamless intercommunication, environment variables, snapshots, rollback, user groups and teams etc etc).

You can go download and try it out today for free (it is open source afterall). Or reach out to us and we can give you a 1:1 demo and have a chat / offer advice?

There are other products out there that do a similar thing but personally, I think FlowForge is a fantastic piece of engineering that will help you with this rather sizable task!

You haven't really identified "big" in this context. 200 devices isn't really that many. At least if you are getting smallish amounts of data once a minute. On the other hand, it might be a lot if you needed to control them real-time with ms updates and data feedback.

With my architecture hat on, I'd say that 1 instance would be the best approach but watching out for performance bottlenecks and having a plan both for business continuity (what happens if your single instance has a problem) and horizontal or vertical scaling should you hit a bottleneck.

Horizontal scaling is adding more, probably low-cost servers (which might be virtual) that work in concert. It also has the advantage of reducing single points of failure but is obviously more complex to set up and orchestrate.

Vertical scaling is replacing kit with ever larger servers.