I need to configure UFL to import data from txt/csv file. My question is: where install it? On PI Server or on PC that will collect the data from txt/csv file.
The best practice is to install interfaces on a stand alone box. It will not compete for resources with the PI Data Historian and you can buffer the data if you need to turn the PI Server off for maintenance without data loss. Well, I guess for UFL, data loss is not typically a concern since you have it in the flat files!
In addition to what Dan shared, here is some information on interface performance that might help you:
one thing that can dramatically affect performance when using ASCII file input is the number of files in the input folder. If thousands of input files exist, performance will be impacted by the number of already processed files since they are located in the same folder as unprocessed files. Each time the interface opens the next file to process, the interface is forced to search through thousands of files to find the next one. To mitigate the impact, you can set the PURGETIME to a small value, such as 1 minute to purge processed files and keep them to a minimum.
Network latency between the PI Data Historian and the interface plays a large role in performance when creating/caching tags. Consider your network latency when deploying the interface.
Retrieving data ...