We have several PI servers throughout the globe. Some of the tags on the local servers are populated to the central PI server using PI-to-PI interface. That was done before AF and has been functioning well.
I need to populate a new sets of tags to the central PI server with proper unit conversion and I'm contemplating 3 options.
1. Use performance equations on local PI servers to create tags (with globally unified UOM) on local PI server, then use PI-to-PI interface to populate central PI server tags. This is how it was done the last time
2. Use analysis on local AF servers to create new tags on the local PI servers, then use PI-to-PI interface to populate central PI server tags
3. Pull in local PI tags data to central AF server, then use AF Analysis to create new PI tags on the central PI server
Option #3 seems to be the easiest as it's a one-step process and we don't need to create PI tags on local PI server that no one would use. But I'm concerned with the network traffic it may create or the load on the central AF server. Another point of consideration is robustness of recovery if there's a network outage.
What's your opinion on this? Someone is probably going to say use PI Cloud Connect. We don't have that now so it's not an option at this point.