I agree that the PI-SDK installation kit sometimes has errors that can be ignored by running the PI-SDK MSI individually. Fortunately, the PI-SDK setup bundle is available individually for download and the MSI can be run directly from that bundle.
Sometimes running an individual MSI with full UI is more forgiving than running that same MSI silently. But it is not clear what numerous problems you are running into.
I am familiar with the reason for this request. The customer does monitoring of the power grid, such as high voltage insulation condition monitoring. They store partial discharge events, where 1 or more discharge event (or burst) can be detected and assigned a single timestamp.
We suggested the following workaround. If a burst of data came in at time X, the data would be stored in PI at X, X+1s, X+2s,X+3s,... and so on up to a max of about 20 or so values. This might work because these bursts of data occur approximately every 500 seconds or so, which means that a grouping of 20 events could be identified as coming from a single burst of events.
While this workaround might affect the way that a trend might appear, the customer is very interested to display the data in XY plots and this workaround might not affect the XY plot in the least. Of course, the XY plot must be done using values from 2 or more different data streams. They have approximately 9 related data streams in each burst.
OK. I fixed AF Server and AF Client. I sent a general email out to setup developers to make this change.
I'm actually not sure what problem you are trying to solve. Certainly there are quirks with installing a newer AF Server with an old PI Data Archive. Why are you bringing up PI Data Archive 2012 in the first place? Why not use PI Data Archive 2016 R2, largely considered the most stable version.
What does "get better economies of scale" mean in this context? Exactly what resource limitations are customers running into that Docker will solve?