-----Original Message----- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of R.S. Sent: Monday, February 05, 2007 11:47 AM To: [email protected] Subject: SIO in SDSF vs read/write speed
I made the following observation: 1. WRITE. I put some data to tape (3592J). The average transfer speed was approx. 70MB/s. SDSF shows very high SIO rate for the job, approx. 2500. Data generated by IEBDG, blocksize=32720, RECFM=FB, LRECL=80 2. READ. I read *the same* data, which I previously wrote. Got even higher transfer speeds, approx. 100MB/s,but significantly lower SIO rate, approx. 40 Dataset (FB 80 32720) read by ICEGENER. I think, ICEGENER reads the data in much larger chunks, so the number of I/O is lower, but the speeds is even greater. Am I right ? <SNIP> First, how big are your buffers? (BLKSIZE/BUFL) Second, how many buffers are you using? (BUFNO) How many can be scheduled at a wack? (NCP) Make everyone do the same and you should get the same results (within reason given load on system, channel utilization, etc.). Later, Steve Thompson ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
