Hi Mieke,  I don't think there is an actual direct limit to the size and usage on (un)saved parameters.  (if there is, I'm also interested to know/learn 😊)

In regards to storing parameters, I came across following topic on Dojo:
string values on a table are truncated - DataMiner Dojo

So it seems we do have a limit for storing data in MySQL databases of 65535 bytes. So depending of your encoding this is a limiter to a fixed size on content that could be stored in MySQL DB's.

Since most systems are no longer using MySQL; Normally element data will get stored in Cassandra and I believe there is no real limit to it anymore on data size.

One important thing to keep in mind though: There is no infinite bucket of space for you to work with.
The bigger the data, the bigger the impact on memory.  The impact might be smaller on a giga server compared to the issues you might face on a mini specs server. But be careful.

The memory cap for DataMiner processes running in 32 bit mode are also limited.  If you have a case where you need to pass around datasets of for example +100MB or gigabytes in content, you probably also want to review your design for internal flow. Transferring such large chunks of data can have a huge footprint on your DMS.

Using C# code to read in a (large) parameter value, means that the full content/size of that parameter that was loaded in the SLProtocol process is now also available in the SLScripting process memory.

If your connector doesn't have the best architecture/flow in place, and you start to duplicate multiple times large data-sets or even store them on displayed parameters (where SLElement would then also load the data in it's memory): the memory limits of your node could become problematic.

But in general, I think for most cases you should be OK to store and work with even large data sets.  🙂