Taking this Sunday as the beginning of the universe:
On 29 May, HSM will dump both volumes to tape#1 from the scratch pool.
On 5 June, HSM will dump both volumes to tape#2 from the scratch pool.
On 12 June, HSM will dump both volumes to tape#3 from the scratch pool.
On 14 June, HSM will release tape#1 back to the scratch pool.
On 19 June, HSM will dump both volumes to some tape from the scratch
pool, possibly tape #1. Assume it does reuse tape#1.
On 21 June, HSM will release tape#2 back to the scratch pool.
On 26 June, HSM will dump both volumes to some tape from the scratch
pool, possibly tape #2. Assume it does reuse tape#2.
...
You only need three tapes to keep HSM running this particular dump class.
That leaves 14 slots in your TS3400 for other uses.
A STACK(nn) operand where nn is greater than the number of volumes in the
dump class provides no additional benefit (and can be confusing). In your
case, each dump tape will never contain more than two volumes.
:>: -----Original Message-----
:>: From: IBM Mainframe Discussion List [mailto:[email protected]] On
:>: Behalf Of Jake anderson
:>: Sent: Friday, May 24, 2013 1:30 AM
:>: To: [email protected]
:>: Subject: Re: HSM option
:>:
:>: Thanks for response. I have read the stack topic, you suggested in
:>: previous
:>: emails. Now my understanding is.
:>:
:>: As we are saving the dumps for only 16 days and most three dump volumes
:>: in
:>: use at any point of time.
:>: So, whenever our DUMP cycle execute on every wednesday at (0300 to
0400),
:>: it will ask for scratch tape volume for dumping the first MIGL00 volume
:>: and
:>: then use same tape volume to dump second MIGL01 volume.
:>: Now when next DUMP cycle execute on next wednesday at (0300 to 0400), it
:>: will again ask for scratch tape becuase
:>:
:>: Never contains more than one dump copy of the same source volume.
:>: DFSMShsm selects empty dump volumes as targets for auto or command dump.
:>:
:>: So, this is how HSM works.
:>:
:>: but as I have mentioned in prevoius email that we are migrating from
:>: ATL3494 to TS3400 tape library which has only 17 I/O slots for data
:>: cartridge, we should always have one tape volume in
:>: scratch status to perform this dumps in weekly basics.
:>:
:>: Please correct, if my understanding is wrong
:>:
:>:
:>: On Fri, May 24, 2013 at 1:22 PM, retired mainframer <
:>: [email protected]> wrote:
:>:
:>: > :>: -----Original Message-----
:>: > :>: From: IBM Mainframe Discussion List [mailto:IBM-
:>: [email protected]]
:>: > On
:>: > :>: Behalf Of Jake anderson
:>: > :>: Sent: Thursday, May 23, 2013 10:28 PM
:>: > :>: To: [email protected]
:>: > :>: Subject: Re: HSM option
:>: > :>:
:>: > :>: Hello,
:>: > :>:
:>: > :>: 1) HSM request for scratch tape, when performing a full volume
:>: dump of
:>: > :>: the
:>: > :>: volumes
:>: > :>: associated with some dump class. We have setup like below
:>: >
:>: > These are obviously not the actual control cards in your ARCCMDxx
:>: member so
:>: > we really cannot tell how accurate they are. You should use cut and
:>: paste
:>: > instead of retyping.
:>: >
:>: > :>: ADDVOL MIGL00
:>: > :>: UNIT(3390)
:>: > :>: MIGRATION
:>: > :>: (MIGRATIONLEVEL1
:>: > :>: AUTODUMP(MIGDMP)
:>: > :>: NOSMALLDATASETPACKING)
:>: > :>: THRESHOLD(85)
:>: > :>: ADDVOL MIGL01
:>: > :>: UNIT(3390)
:>: > :>: MIGRATION
:>: > :>: (MIGRATIONLEVEL1
:>: > :>: AUTODUMP(MIGDMP)
:>: > :>: NOSMALLDATASETPACKING)
:>: > :>: THRESHOLD(85)
:>: > :>:
:>: > :>:
:>: > :>: As per above code, we have two migration volumes (MIGL00 and
:>: MIGL01)
:>: > :>: defined under MIGDMP dump class.
:>: > :>:
:>: > :>:
:>: > :>: 2) There are only two DASD volume associated with this dump class.
:>: > :>:
:>: > :>: 3) In my previous email, I have wrongly put the dump time 0500 to
:>: 0600.
:>: > :>: It
:>: > :>: is actually dumping MIGL00 and MIGL01 volume on every Wednesday
:>: between
:>: > :>: 0300 to 0400.
:>: > :>:
:>: > :>:
:>: > :>: In our case, when every Wednesday dump starts at 0300, it ask for
:>: > :>: scratch
:>: > :>: tape volume, but I actually want dump of MIGL00 and MIGL01
:>: volumes
:>: > :>: going
:>: > :>: into the used tape rather then scratch tape.
:>: >
:>: > The HSM volume dump function will not do this. Go back and read the
:>: quotes
:>: > from the manual I included in my previous message (and left again for
:>: you
:>: > at
:>: > the end of this one). Then go and read the full description in the
:>: manual.
:>: >
:>: >
:>: > :>:
:>: > :>: I want to do this because, we are planning to migrate from ATL3494
:>: to
:>: > :>: TS3400 tape library, which has very minimum number of I/O slots
:>: where
:>: > we
:>: > :>: can keep data cartridge as compare to ATL 3494.
:>: > :>:
:>: > :>: if HSM ask for scratch tape volume everytime , then I will shortly
:>: run
:>: > :>: of
:>: > :>: the tapes and dump will fail. So, I want HSM to use the tape
:>: volume
:>: > :>: which
:>: > :>: has valid data and fill it to maximum before it
:>: > :>: ask for new scratch tape.
:>: >
:>: > HSM will never place a second dump of a DASD volume on the same tape
:>: volume
:>: > that already holds a previous dump of that DASD volume.
:>: >
:>: > However, since you save the dumps for only 16 days, at any point in
:>: time
:>: > you
:>: > will only have at most three dump volumes in use. On the fourth
:>: weekly
:>: > execution, the tape from the first will already be scratched and you
:>: can
:>: > reuse it.
:>: >
:>: > :>: On Thu, May 23, 2013 at 11:36 PM, retired mainframer <
:>: > :>: [email protected]> wrote:
:>: > :>:
:>: > <snip>
:>: > :>: > If you look at the description of the STACK operand in the HSM
:>: > Storage
:>: > :>: > Administration manual (Chapter 36 in my 1.11 copy), you will
:>: find:
:>: > :>: >
:>: > :>: > "DFSMShsm selects empty dump volumes as targets for auto or
:>: command
:>: > :>: dump."
:>: > :>: >
:>: > :>: > and
:>: > :>: >
:>: > :>: > "A given dump volume:
:>: > :>: > ...
:>: > :>: > Never contains more than one dump copy of the same source
:>: > volume.
:>: > :>: > ..."
:>: >
:>: > ----------------------------------------------------------------------
:>: > For IBM-MAIN subscribe / signoff / archive access instructions,
:>: > send email to [email protected] with the message: INFO IBM-MAIN
:>: >
:>:
:>: ----------------------------------------------------------------------
:>: For IBM-MAIN subscribe / signoff / archive access instructions,
:>: send email to [email protected] with the message: INFO IBM-MAIN
----------------------------------------------------------------------
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to [email protected] with the message: INFO IBM-MAIN
