Buffering 54 Primary and Secondary Memory Buffer Size (Bytes) This is a key parameter for buffering performance. The sum of these two memory buffer sizes must be large enough to accommodate the data that an interface collects during a single scan. Atypical event with a Float point type requires about 25 bytes. If an interface writes data to 5,000 points, it can potentially send 125,000 bytes (25 * 5000) of data in one scan.
As a result, the size of each memory buffer should be 62,500 bytes. The default value of these memory buffers is 32,768 bytes. OSIsoft recommends that these two memory buffer sizes should be increased to the maximum of 2000000 for the best buffering performance.
Send rate (milliseconds) Send rate is the time in milliseconds that Bufserv waits between sending up to the
Maximum transfer objects (described below) to the PI Server. The default value is 100. The valid range is 0 to 2,000,000.
Maximum transfer objects Max transfer objects is the maximum number of events that
Bufserv sends between each Send rate pause. The default value is 500. The valid range is 1 to 2,000,000.
Buffered Servers The
Buffered Servers section allows you to define the PI Servers or PI collective that the buffering application writes data.
PIBufss PIBufss buffers data only to a single PI Server or a PI collective. Select the PI Server
or the PI collective from the Buffering to collective/server drop down list box. The following screen shows that PIBufss is configured to write data to a standalone PI Server named starlight. Notice that the
Replicate data to all collective member nodes checkbox is disabled because this PI Server is not part of a collective. (PIBufss automatically detects whether a PI Server is part of a collective)
