Hi All,
Do you have any measures for the relative performance of the jZOS I/O routines relative to the standard JRIO routines? Question
We are using JRIO at the moment and it is very painful reading/writing zOS datsets relative to the standard JAVA I/O for HFS/ZFS files.
Many thanks
Steve
Perfmance of jZOS I/O Routines relative to JRIO ?
-
SteveBagshaw
- Posts: 4
- Joined: Thu Dec 30, 2004 7:16 pm
- Location: UK
Sorry, we don't have any comparative data; maybe someone else has done this and will share....
We have found that it is -extremely- important to put "noseek" on the openmode for sequential datasets when opening in type=record mode.
For example:
new ZFile("//DD:SYSUT1", "rb,type=record,noseek");
new ZFile("//DD:SYSUT2", "wb,type=record,noseek");
Given this, I have found that the ZFile I/O routines seem to perform very well. I have not compared to JRIO, though. JRIO, as you say, is painful.
If you do any benchmarks comparing them, it would be great if you could post yor results.
We have found that it is -extremely- important to put "noseek" on the openmode for sequential datasets when opening in type=record mode.
For example:
new ZFile("//DD:SYSUT1", "rb,type=record,noseek");
new ZFile("//DD:SYSUT2", "wb,type=record,noseek");
Given this, I have found that the ZFile I/O routines seem to perform very well. I have not compared to JRIO, though. JRIO, as you say, is painful.
If you do any benchmarks comparing them, it would be great if you could post yor results.