At least on modern tape drives 16K is not the actual physical block size on tape, so it doesn't cause inefficient media usage. Ever since automatic hardware compression has been forced, the actual tape media physical blocks have been "superblocks" with a size of the controller's choosing, transparent to the operating system. The smaller block size used by z/OS requires more buffer management overhead and more channel commands for the data transfer, but at least it doesn't kill physical space on the media. Not like the old days where a singularly bad choice of block size could easily waste 99% of the media.
     JC Ewing

On 02/07/2013 08:39 AM, David Devine wrote:
Hi Alain,
It is not currently possible to change the 16k blksize on Hsm migration or 
backup tapes.
Regards
              Dave

***********************************************************************************************
  Hello,
Thank's for your update, but do you an idea to change the value, just because I'm sure that the default value is not use
Regards
Alain -----Message d'origine-----
De : IBM Mainframe Discussion List [mailto:[email protected]] De la part 
de David Devine
Envoyé : jeudi 7 février 2013 12:24
À : [email protected]
Objet : Re: blksize for ML2 migrated dataset via HSM
Hello,
The Datamover parameter of DSS or HSM only applies to Control dataset backups 
and is not for general usage.
Hsm has used a 16k blocksize for backup and migration tapes since inception, possibly because originally it was a good performance match for the specs of existing 3420 "polo" tape drives and the upcoming 3480 "cartridge" tape drives.
(People who read the announcment specs in the early 80's feel free to dive in!)
Well overdue for an update. Dumps however are straight dfdss and depending on what Z/os release you are on, the blksize will be the default of 256K (Zos 1.12 up) or 64K (Zos 1.11 down) or even 32k if you use the patch! While we are at it, an ML3 level for long term archive datasets (greater than 5 years say) would be good to split them out from all the other stuff on ML2. regards,
              Dave
*********************************************************************************************** Hello, I observed a blksize of 16k for all ml2 migrated dataset (seen on the CA1 product). I use DSS as datamover and i'm wondering about the size concerning the blksize. Why dss don't use 256k since zos 1.12 ?
Everybody will have an answer. thank's

--
Joel C. Ewing,    Bentonville, AR       [email protected] 

----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN

Reply via email to