Why does CA recommend the region size for the TPX proc be set to 0M?
The reason why we recommend changing the REGION size to 0M is because TPX will get the amount of storage required and then TPX will return the rest of the storage to the system. With the TPX region defined as REGION=0M, there is no storage limitations.
In the start phase TPX issues a conditional GETMAIN asking the operating system how many storage is available below the line.
Assume the operating returns: We have 8.5 MB available.
If TPX would do a GETMAIN now for 8.5 MegaBytes there would be no storage left and any GETMAIN from e.g. operating system components would fail with S80A, S878 abends. TPX would not come up. So TPX checks the 'System Storage' value and subtracts this from the 8.5 MB available.
If 'System Storage' is set to 2048K TPX does a GETMAIN only for 6.5 MegaBytes for its modules, Slots and DSA below.
The 2 MB left remains under the control of the operating system to satisfy GETMAIN requests below the line.
TPX itself does not do any explicit GETMAIN here unless you need to do a VTADD. For every VT added with VTADD TPX will request round about 150 bytes (actually 144) above the 16M line through a GETMAIN.