gtpm1m2h | TPF V4R1 Migration Guide: 3.1 to 4.1 |
To protect filed data from being overlaid, in the TPF 4.1 system the data is copied from the user core block into a virtual file access (VFA) buffer whether a candidate or not. As a result, VFA is active in all system states, with VFA delay file capability available only in NORM state. Noncandidates are flushed after they are filed.
Programs are run directly from VFA, improving the performance of file resident program enters.
New online and offline VFA performance monitoring facilities allow better tuning of VFA candidates and VFA resources.
There are also enhancements to the ZVFAC command that allow records to be flushed or purged from VFA by file address or record ID.
All DASD FIND and FILE macros, except the FDCTC macro, now interface with VFA.
There is a new interface from exception recording to VFA that allows VFA to leave the delay file attribute enabled while exception recording ends.
IBM recommends that you use the virtual file access (VFA) package and that you use it for a while before tuning the following two new indicators:
The processing of a FINSC or FILSC macro forces a flush of the VFA buffer if the record is currently in VFA.
IBM recommends increasing the working storage by a factor of 3 to 5 and increasing the virtual file access (VFA) storage by 20 percent.
For example, if you have 10 MB allocated for working storage and 60 MB allocated for VFA storage, increasing working storage by a factor of 3 to 5 and increasing the VFA storage by 20 percent would cause your working storage to grow to 30 to 50 MB and VFA storage to grow to 72 MB.