|
|
| Author |
Message |
sravani
New User
Joined: 17 Jul 2005
Posts: 3
|
|
|
|
Hi,
My JCL has the following statements. Can anyone tell me why BUFNO is used for output datasets?
//SORTOUT DD DSN=&&TEMP1,BUFNO=29,
// DISP=(NEW,PASS,DELETE),
// SPACE=(CYL,(1500,1000),RLSE,,ROUND),
Thanks,
Sravani |
|
| Back to top |
|
 |
References
|
|
 |
withnams
New User
Joined: 06 Jul 2005
Posts: 26
Location: Chennai
|
|
|
|
| It perhaps means the number of buffers alloacted for that dataset. |
|
| Back to top |
|
|
ironmike
New User
Joined: 07 Aug 2005
Posts: 37
|
|
|
|
BUFNO= is the number of buffers that will be used for buffering output for that data set. The example you coded appears to be for a sequential file. Increasing the number of buffers can improve performance if QSAM is used for writing to an output sequential file. Your example is for a SORTOUT DD, so the program writing to the file may be the SORT; SORT probably used BSAM or EXCP to write to this file, in which case SORT is probably doing intelligent buffer management and the BUFNO=29 may be ignored; if your program and not the SORT is writing to the file out of a SORT exit (e.g., the SORT E35 exit), then your program may be using QSAM and bumping up BUFNO can speed up the output process.
Here's a quote from IBM's DFSMS Using Data Sets manual:
| Quote: |
| In QSAM, the value of BUFNO determines how many buffers will be chained together before I/O is initiated...When enough buffers are available for reading ahead or writing behind, QSAM attempts to read or write those buffers in successive revolutions of the disk. |
Default BUFNO for a sequential, non-SYSOUT file is five.
Basically, if you've got no virtual storage constraints, large BUFNO = goodness!  |
|
| Back to top |
|
|
mmwife
Super Moderator
Joined: 30 May 2003
Posts: 1526
|
|
|
|
DFSORT and Syncsort calculate and use their own BUFNO params for I/P and O/P.
Those supplied in the JCL are ignored. |
|
| Back to top |
|
|
|
|
|
