DIV (Data in Virtual) is an access method which differs from the traditional access methods such as ISAM, SAM or UPAM in that it functions without structuring the file in records or blocks, without I/O buffer and without special I/O macros (e.g. GET, PUT).
DIV processes only PAM files of the type NK4 (BLKCTRL=NO, the block size being a multiple of 4 KB).
DIV regards the file as a linear sequence of bytes. The DIV function MAP enables a file or a file area to be assigned to an area in the virtual address space. The virtual address space assigned to a file area then forms a "window" in which the pages of the file appear automatically when the customary CPU commands are used to access the corresponding pages in the virtual address space. Data modified in one window can be written to the file using the DIV function SAVE.
In the steady state the file or file area resides in the user's address space and is thus subject to paging. When the file areas are large and insufficient main memory is available, the paging rate increases correspondingly.
Performance is increased when data in the window which was read into the window by preceding accesses (via paging) is accessed repeatedly.
DIV reading
The maximum performance enhancement occurs when data in the window is accessed; no disk access is required.
If data access needs to be handled using page transfer, the path length per access is approx. 20% shorter than with read access using UPAM.
DIV writing
Secure writing is implemented by writing the modified page back to the file after every write access (DIV-SAVE). In this case the path length per access is approx. 35% longer than with a UPAM write access. The extra effort can only be reduced by decreasing the number of SAVE-IOs (e.g. saving the file once only, when CLOSE is issued). If page transfer is required for write access, the path length is approx. 10% longer than with UPAM write access.
The access method has a great influence on the performance of the entire system and should consequently be selected with care.