Node red - add-on or separate

Have set up my Proxmox system on NUC with home assistant installed as per the instructions on here.

Currently added Node Red into it - but having used Node Red before I expect I will use it to script my more complex flows, so is there any advantage, or disadvantage, in setting it up to run in its own VM? Or is that unnecessary overhead, since the integrated version works well. My concern is that if I bring in a lot of external packages into it, it’ll make my HA instance run more slowly…

ETA: will run Ubuntu in another VM, with Docker, so node red, Deepstack, various motion and recognitions systems, and some other machine learning stuff all in containers in there. Or keep Node Red inside the home assistant vm instance?

One way to look at this question is that NodeRED will require resources (processing power, memory, storage) regardless of how it’s added.
If you set up an additional VM, you’ll need to assign it a few cores or threads and some memory and its own storage space. You’ll have some overhead required to run the VM as well. Those resources have to come from somewhere. Either you have some cores/threads, memory, and storage you weren’t already using in your VMs or you’ll be taking them from the resources of the other VMs. In the later case, you really don’t gain anything and may even lose a bit.
If you keep NR & HA together you could assign those resources you would have used in the additional VM to the HA VM (or at least not borrow from it to create the additional VM).
So, unless you have resources, particularly processing power, that you’re not already using, it would be more efficient to keep NR and HA together. If, on the other hand, you have a few cores that are just lying dormant, a separate VM could provide more performance. (I don’t have any hard metrics on just how much that might be though.)

1 Like

fair enough. Seems like there’s not much overhead in having it use RAM in the HA instance then. Might still separate it off as it would then be easier to migrate to aniothe machine later if I needed more horsepower - will look into how easily it integrates across VMs…