Are there any documents or tips for what a PI Administrator should do to assist and support AF and the effects of what users create in AF could have on the PI Data Archive server, such as performance?
I would start off by creating and monitoring some performance coutners on the PI Data Archive. If performance starts to suffer, then you can investigate some more. Note that PI Analyses will have the greatest impact on the PI Data Archive. For Analysis Service, I would suggest you to take a look at https://techsupport.osisoft.com/Troubleshooting/KB/KB01312/ and Troubleshooting PI Analysis Service Performance Issues: High Latency.
Concerning the PI Data Archive itself, I would start by monitoring the following performance counters (they can be created with SMT -> IT Points -> Perf Counters):
System Up Time is the elapsed time (in seconds) that the computer has been running since it was last started. If this tag goes bad or stale it could indicate a problem with the OS.
% Processor Time is the percentage of elapsed time that the processor spends to execute a non-Idle thread. If this tag is pegged at 100% for extended period of time, you would need to investigate what is causing it
Available MBytes is the amount of physical memory, in Megabytes, immediately available for allocation to a process or for system use.
Free Megabytes displays the unallocated space, in Megabytes, on the PI Server's disk drive. Free up space if this falls below 500 MB, the size of your event queue, or the size of an archive.
Elapsed Time is the total elapsed time, in seconds, that this process has been running. If these go bad or stale, make sure that the process is running and not in error
Events in Overflow Queues is the total number of events in the overflow queue files. This value should be zero when the PI Server is processing incoming data properly. When this value starts growing there could be a problem with archive corruption, event queue corruption, or another issue preventing the archive subsystem from processing new data.
Monitoring and saving the history of the main performance counters is a great way to keep your PI System healthy and find the causes of the issue when something goes wrong.
Retrieving data ...