What is everyone recommending for minimum PC specs? I know you can get away with going pretty low, depending on project size and deployment, but what's the lowest you'll actually go?
Personally, I don't feel like I could recommend any less than an i3-13100, 16 GB RAM, and an SSD. Even still I would prefer at least an i5-13500 (just personal experience with PCs).
Ignition will typically be running on Windows. Running as the gateway, possibly as a client at the same time.
In this specific instance we're replacing a PC with an i7-7700 but a modern i7 feels like way overkill. The i3 is already at least a 40% improvement.
I suppose it depends, are we talking about a client or a gateway? Vision or Perspective?
We use WYSE thin clients for our Vision Clients (they all remote into a RDS Host), and they have Intel Celeron processers @1.5GHz, with 4GB ram. Never had an issue with them personally. They're more than enough for a perspective client, IMO.
Re vision I have found similar as @lrose . With best practices you can get away with the relatively weak thin clients.
It's really hard to say in a complete vacuum though. 16 gigs is probably overkill for 1 HMI style project and probably no where near enough for perspective projects with 100+ users.
Not to mention, how many devices will be connected, and if you're using the builtin-tag history or rolling your own. How many tags will the project be using, are you utilizing auditing, alrams, etc...
So basically it depends is always the answer. That's quite annoying.
IA should really have a hardware recommendation tool where you go through and fill out a form with the use case info and it spits out minimum requirements. I find the guidelines that they have currently very unhelpful, hence my question here.
Yeah, but the platform is so open and flexible that the form would have to be extremely detailed. And how could IA account for the crappy binding and coding practices that, themselves, will crush performance?
Get some various levels of hardware for your lab. Test your application on them.
I mean, it couldn't be any worse than expecting those same people to pick from the arbitrary "Small," "Medium," and "Large" projects here.
I'd imagine something like this could be doable.
Form -> best case scenario minimum hardware + 35% buffer for user error = recommended hardware.
Actual hardware would also be better than "4 Cores*, 8GB memory, SSD." What 4 cores are we talking about? i3-14100 or X4 970? What RAM are we talking about DDR3, DDR4, DDR5?
I don't mean this in a derogatory way in case it comes across this way in text, but have you seen a lot of OTHER ignition projects that aren't yours?
Being open and flexible is the double edged sword with Ignition. You would be surprised at how wildly different some other people executed SCADA ideas compared to you, and hence the performance load could be wildly different from one project to another.
It really is impossible for IA or anyone to ballpark performance metrics. When people ask me this, and are frustrated by the "hard to say" answer - the best answer is to self reflect because all they think about is their own SCADA design. If you are wanting to know how YOUR Ignition SCADA design can scale, then test your own design. Ignition's trial structure makes it really easy to do that.
