How much better would performance be using PI AF attributes with <None> for data reference, then using a service to update the values and timestamps that I can gather from external data sources? For example, I have 40 attributes that use the Lookup Table data reference, and SQL that gets the attribute value from an internal PI AF table. When we do searches in client tools, the AF client(s) on whichever application seems to want to load a lot more elements/attributes/values/timestamps then we are looking for. Does the action of Lookup Tables take more time to execute the SQL statements on each attribute value, than just simply returning the current value of an AF attribute/timestamp which is basically static? Take for example a list of 5000 well assets, with references to at least two different element hierarchies. Coresight 2016 still seems to want to load all the assets, including the references, each time a user drills down into the parent element. Would the loading be a lot faster if the attributes were all either PI Point or simulating static values using the <None> data reference?