Sam, At the very least you should have coded OPT(4). That way you would have been reading 1 cylinder at a time vs. 1 track. The default is OPT(1) which is one track processing.
Personally I would have altered the Mgt. class to something with a reasonable migrate/expire criteria and let HSM handle it. Just curious, what is your Setsys Dumpio set to? Regards, Dave O'Brien -----Original Message----- From: Knutson, Sam [mailto:[EMAIL PROTECTED] Sent: Tuesday, June 13, 2006 9:18 AM To: [email protected] Subject: DFSMSdss Performance Question Hi, Is there a way to convince DFDSS to dump from multiple target volumes in any simple way where DFDSS picks the volumes to a single output tape to speed things up? I looked through the Reference Guide and nothing jumped out. I have FDR/FDRDSF so Bruce feel free to chime in with a comparison if you think I could have done better. I needed to backup 1400 SVC dumps (compressed PS-E) total about 20 million tracks across 276 volumes and then I deleted all but two. I didn't know if I might later need one of the many and the outage had impacted our highest availability LPAR so I didn't want to throw any doc away. DAE wasn't doing much for these abends and while I had already worked with the first and last dumps I knew I might need to revisit something that happened in the middle. I thought about just doing an HMIGRATE my.svc.dump.date.** to ML2 but figured it would be better to dump all this directly onto a few high capacity tapes I could scratch when I was sure I didn't need them. The UNIT and DATACLAS put this to 3592 tapes so I only needed 3 output tapes. //DFDSSCPY EXEC PGM=ADRDSSU //SYSPRINT DD SYSOUT=* //TO1 DD DISP=(NEW,CATLG,DELETE), // UNIT=TAP18,DATACLAS=T3592L1,VOL=(,,,59), // DSN=DSS.ASYS.SDUMP.D060611.BACKUP,LABEL=RETPD=365 //SYSIN DD * DUMP DATASET( - INCLUDE(DSS.%SYS.*.D060611.**)) - ALLDATA(*) - OUTDDNAME(TO1) This took about 742 minutes (all night) and processed each volume on which data sets were found sequentially. Is there anything I can do with DFDSS that would have significantly improved the elapsed time? Any way to automatically process multiple input volumes at the same time? The key is I don't want to code a list of volumes by hand that was not feasible in this case and most of the time it isn't. I guess I could write a little REXX using CSI to build DFDSS JCL but I would prefer the dump utility handle the parallelism itself. Best Regards, Sam Knutson, GEICO Performance and Availability Management mailto:[EMAIL PROTECTED] (office) 301.986.3574 There's nothing new under the sun, but there are lots of old things we don't know. Ambrose Bierce ==================== This email/fax message is for the sole use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution of this email/fax is prohibited. If you are not the intended recipient, please destroy all paper and electronic copies of the original message. ---------------------------------------------------------------------- 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 ---------------------------------------------------------------------- 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
