z/OS DFSMS Using Data Sets
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Managing SAM Buffer Space

z/OS DFSMS Using Data Sets
SC23-6855-00

The operating system provides several methods of buffer acquisition and control. Each buffer (virtual storage area used for intermediate storage of I/O data) usually corresponds in length to the size of a block in the data set being processed.

You can assign more than one buffer to a data set by associating the buffer with a buffer pool. A buffer pool must be constructed in a virtual storage area allocated for a given number of buffers of a given length.

The number of buffers you assign to a data set should be a trade-off against the frequency with which you refer to each buffer. A buffer that is not referred to for a fairly long period could be paged out. If much of this were allowed, throughput could decrease.

Using QSAM, buffer segments and buffers within the buffer pool are controlled automatically by the system. However, you can notify the system that you are finished processing the data in a buffer by issuing a release (RELSE) macro for input, or a truncate (TRUNC) macro for output. This simple buffering technique can be used to process a sequential data set. IBM recommends not using the RELSE or QSAM TRUNC macros because they can cause your program to become dependent on the size of each block.

When using QSAM to process tape blocks larger than 32␠760 bytes, you must let the system build the buffer pool automatically during OPEN. The macros GETPOOL, BUILD, and BUILDRCD do not support the large block size or buffer size. If, during QSAM OPEN, or a BSAM OPEN with a nonzero BUFNO the system finds that the DCB has a buffer pool, and that the buffer length is smaller than the data set block size, an ABEND 013 is issued.

For QSAM, IBM recommends that you let the system build the buffer pool automatically during OPEN and omit the BUFL parameter. This simplifies your program. It permits concatenation of data sets in any order of block size. If you code RMODE31=BUFF on the DCBE macro, the system attempts to get buffers above the line.

When you use BSAM or BPAM, OPEN builds a buffer pool only if you code a nonzero value for BUFNO. OPEN issues ABEND 013-4C if BUFL is nonzero and is less than BLKSIZE in the DCB or DCBE, depending on whether you are using LBI. If the system builds the buffer pool for a BSAM user, the buffer pool resides below the 16 MB line.

If you use the basic access methods, you can use buffers as work areas rather than as intermediate storage areas. You can control the buffers in a buffer pool directly by using the GETBUF and FREEBUF macros.

For BSAM, IBM recommends that you allocate data areas or buffers through GETMAIN, STORAGE, or CPOOL macros and not through BUILD, GETPOOL, or by the system during OPEN. Allocated areas can be above the line. Areas that you allocate can be better integrated with your other areas.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014