we have PI collective(3 PI servers) and PI AF server,
is there any point of enabling buffering on PI AF server, at the moment we have analysis reading data from PI collective and sending back after calculation.
I would highly recommend buffering your analysis outputs on the AF Server. This will allow the output data to be fanned to each collective member and will aid in data retention in case of network disruptions. You can follow to procedure below to buffer Analysis Service outputs:
can you give more detail on this "This will allow the output data to be fanned to each collective member and will aid in data retention in case of network disruptions", some example?
If the Analysis Service outputs are not buffered, the output data will only be written to the collective member to which it is connected (typically the primary server). The data will only be archived on one Data Archive server and will only be able to be queried from that same server. With buffering enabled, the data will be replicated and archived on all three collective members. Additionally, if the connection between the Analysis Service and Data Archive is interrupted, data will be held in the buffer and written to the archive once the connection is re-established. This behavior is very similar to the way buffering for interfaces works.
Hope this helps.
Retrieving data ...