I'm trying to sort out what circumstances would lead me to select one tag historization sample mode over another. I definitely might be missing things here, so I'm looking forward to any insights the forum might provide!
-
If I want to guarantee that I check a value at least once every period (say, once per second), I can do this with OnChange and Max Time Between Samples.
-
If I want to make sure I capture all changes, I can do this with OnChange and a deadband of zero.
-
If I want to capture when multiple changes occur between standard sampling times, I can only do this with OnChange - and I can limit it with Min Time Between Samples, as well as deadband settings.
(I'd love to have more info on this last point too though, if anyone has more details - I'm not sure of the underlying mechanisms of how OnChange Min Time interacts with Tag Group Rate and Publishing Interval, regarding how frequently the historian can actually log updates from underlying data changes and server updates. Basically, does anyone know how OnChange data updates/triggers are actually handled?)
The main thing I see that's exclusive to Periodic and Tag Group sampling modes is that these modes use the gateway timestamp, whereas On Change uses the value timestamp (though I only found this in a forum post, not in documentation). Value timestamp seems more frequently "better" to me, but I can imagine scenarios where the gateway timestamp might be preferred. (e.g. to help "synchronize" field device data despite drifting clocks)
More broadly though, I don't see many cases where the On Change mode is not superior to Periodic or Tag Group? I suppose that administering a large number of tags' historization settings via a single Tag Group is convenient, but this isn't really a technical advantage?
Are there any performance advantages (or other technical advantages) to Periodic or Tag Group sampling over On Change [given similar overall numbers of historized data points], and if yes what factors influence those differences/advantages?
Thanks!