Thank you for the numbers.

In a "normal" situation, that is, receiving 500 messages with simple data per second from a handful of servers, this would be perfectly OK. And it might be OK in your case as well, we just do not know. The factors that make it different are:
- the fact that there is so many devices.
- the fact that you mentioned you subscribe to "complex data nodes". Can you elaborate on that? Are they custom data types (such as structures - but represented as single node)? How big they are?

Both approaches (with single EasyUAClient instance, and with multiple EasyUAInstances) are possible, and have their pros and cons. I would probably start with single EasyUAClient instance and see how that works out. The differences are:

- Single EasyUAClient instance: Less memory. More limited in terms of threading - all notifications will be delivered to you from/on a single thread, so if you have multiple cores, it won't - without extra coding - take advantage of that, for the part of your code that processes the notifications.
- Multiple EasyUAClient instance: More memory (and threads). Notifications will be delivered essentially in parallel on these instances, making it possible to process them more efficiently on multiple cores (but may have to synchronize them yourself if they access some common data body).

Best regards

Hello,
thank you for your interest in our products,

I admit that we have not tested with such large number of devices. I encourage you to verify it before purchase. I will send you an evaluation license (by email), valid for 30 days (can be extended if needed), with no run-time limit.

We do not have "hard-coded" limits for the number of devices etc., but your scenario will certainly be memory intensive and possibly also CPU intensive. I do not quite get how the numbers you have provided add up: If you had 1000 devices and each gave 100 messages per day, that would come slightly above 1 message per second on average, which seems reasonable (note: one should also think about how the messages tend to be grouped, time-wise). How does the node with 15-20 changes per minute fit with the fact "100 messages per day"? - this is the part I do not get. If you had 1000 nodes like this, that would come to 333 messages per second. In normal scenarios (with 1 server or a handful or servers), this were quite doable as well, but needs to be tested, it is fairly different from 1 message/second which comes out of the first calculation. Can you please explain where I am getting it wrong.

Best regards