Re: Need MASS recall advice
On Fri, 21 Mar 2008 10:25:01 -0500, Rugen, Len [EMAIL PROTECTED] wrote: I'm doing some planning for a end-of-life for the mainframe conversion project, as in the 9th year of our 5 year plan to get off this system. We have View Direct / Mobius report viewer. It has 100,000's of HSM migrated files in our 3494 library. If they just issue recalls for these in some order they choose, the physical tape and file order will be random and slow. On a good day, 500 recalls would probably be the max we could allow before we would need the drives for overnight production. I think when I did the math, it was 3-4 years worth of recalls. (That's OK, I need this job :-) ) Is there a faster way? If I had a list of files they wanted, could I send the recalls in tape - file sequence somehow to batch them? Thanks Len Rugen -- Hi Len, your scenario practically describes what we've been going through as well. We have been trying to get off of Mobius Viewdirect for about 5 years now but there are some stubborn users who have done everything they can do to prevent this from happening. I have gotten the report archives to around 60,000 records and I basically did the same thing that someone already mentioned which was the TSO HLIST LEVEL(whatever.level.qualifier) OUTDATASET(output.dataset.name) recall routine. I changed the management class to not migrate the archives anymore since there were very few of them so now they just stay on L0. Just out of curiosity, what did you replace VD with? Regards, Gil. -- 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
Re: Need MASS recall advice
On Fri, 21 Mar 2008 10:25:01 -0500, Rugen, Len [EMAIL PROTECTED] wrote: We have View Direct / Mobius report viewer. It has 100,000's of HSM migrated files in our 3494 library. If they just issue recalls for these in some order they choose, the physical tape and file order will be random and slow. On a good day, 500 recalls would probably be the max we could allow before we would need the drives for overnight production. I think when I did the math, it was 3-4 years worth of recalls. (That's OK, I need this job :-) ) Not that this is similar but I had to delete about 500K+ old datasets from a HSM system back when and could generate the delete records to do it. Submitting them within TSO in batch was a pain to do in chucks of a few hundred. I have a TSO DELAY command (on CBT tape or available if you need it offlist). I just inserted the DELAY 1 (delay for 1 second) command in the stream after a number of delete commands and let it rip. It would run for 8-12 hours at a crack for the number I wanted. Indeed getting the recalls coming from the same tape all together is a stellar idea. jim -- 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
Re: Need MASS recall advice
Hi Len. If you submit a bunch of recalls HSM will automatically sort them based on the volser they are on - in other words all of the recalls in the queue, that are on the same volume, will get recalled one after the other. If you use the earlier response of issuing the TSO HLIST LEVEL(whatever.level.qualifier) OUTDATASET(output.dataset.name), you can sort by the physical volser, build the recall commands, submit them to HSM and HSM will get them ordered for you. If you are not using the CRQ then you will need to issue all of the recall commands, for datasets on the same tape, on the same system for HSM to handle it properly - otherwise you will get multiple HSM's contending for the same tape. Hope this helps! ThanksRick -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Rugen, Len Sent: Friday, March 21, 2008 10:25 AM To: IBM-MAIN@bama.ua.edu Subject: Need MASS recall advice I'm doing some planning for a end-of-life for the mainframe conversion project, as in the 9th year of our 5 year plan to get off this system. We have View Direct / Mobius report viewer. It has 100,000's of HSM migrated files in our 3494 library. If they just issue recalls for these in some order they choose, the physical tape and file order will be random and slow. On a good day, 500 recalls would probably be the max we could allow before we would need the drives for overnight production. I think when I did the math, it was 3-4 years worth of recalls. (That's OK, I need this job :-) ) Is there a faster way? If I had a list of files they wanted, could I send the recalls in tape - file sequence somehow to batch them? Thanks Len Rugen -- 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
Re: Need MASS recall advice
Have the CIO key the names into the JCL. I caught you said 100,000's of files. So write a program to read the list and build what your solution method is. Either a job with IEFBR14 or if you used REXX to read the list you could issue the HRECALL command, not caring if it worked or not, you could use LISTDSI with an option to wait and then you know if it worked or not. That REXX code could run IKJEFT1B as a batch job so you do not tie up your TSO session. Letting REXX submit 500+ HRECALL commands will flood HSM but you can get better throughput as multiple recalls from the same tape requested in the right sequence will just go from file to file. TSO allocate may dismount the tape before your next allocation, and certainly the tape will be rewinding. Try batching requests by the tape they are on in the sequence they are on the tape. On Fri, 21 Mar 2008 17:23:09 -0500, Bruno Sugliani [EMAIL PROTECTED] wrote: On Fri, 21 Mar 2008 10:25:01 -0500, Rugen, Len [EMAIL PROTECTED] wrote: Is there a faster way? If I had a list of files they wanted, could I send the recalls in tape - file sequence somehow to batch them? AFAIK , you do not even need to do explicit Hrecall If you have the list , you just run a BR14 and it will trigger an implicit hrecall . The list sorted by volume , in order to reduce the number of mount should improve performance . Bruno Bruno(dot)sugliani(at)groupemornay(dot)asso(dot)fr -- 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
Need MASS recall advice
I'm doing some planning for a end-of-life for the mainframe conversion project, as in the 9th year of our 5 year plan to get off this system. We have View Direct / Mobius report viewer. It has 100,000's of HSM migrated files in our 3494 library. If they just issue recalls for these in some order they choose, the physical tape and file order will be random and slow. On a good day, 500 recalls would probably be the max we could allow before we would need the drives for overnight production. I think when I did the math, it was 3-4 years worth of recalls. (That's OK, I need this job :-) ) Is there a faster way? If I had a list of files they wanted, could I send the recalls in tape - file sequence somehow to batch them? Thanks Len Rugen -- 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
Re: Need MASS recall advice
On Fri, 21 Mar 2008 10:25:01 -0500, Rugen, Len [EMAIL PROTECTED] wrote: I'm doing some planning for a end-of-life for the mainframe conversion project, as in the 9th year of our 5 year plan to get off this system. If I had a list of files they wanted, could I send the recalls in tape - file sequence somehow to batch them? Len, you can use the following command to list level qualifiers to an output file and sort them by volume and use it as input to HRECALL them. TSO HLIST LEVEL(whatever.level.qualifier) OUTDATASET(output.dataset.name) HTH, Pat L. -- 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
Re: Need MASS recall advice
On Fri, 21 Mar 2008 10:25:01 -0500, Rugen, Len [EMAIL PROTECTED] wrote: I'm doing some planning for a end-of-life for the mainframe conversion project, as in the 9th year of our 5 year plan to get off this system. We have View Direct / Mobius report viewer. It has 100,000's of HSM migrated files in our 3494 library. If they just issue recalls for these in some order they choose, the physical tape and file order will be random and slow. On a good day, 500 recalls would probably be the max we could allow before we would need the drives for overnight production. I think when I did the math, it was 3-4 years worth of recalls. (That's OK, I need this job :-) ) Is there a faster way? If I had a list of files they wanted, could I send the recalls in tape - file sequence somehow to batch them? Thanks Len Rugen -- 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 Possibly the LIST TTOC command may provide an information source you can sort, manipulate and generate HRECALL commands, using REXX, DFSORT/SYNCSORT? Obviously, a straight HRECALL with a LEVEL parameter is a comparison test to perform, which should not be used -- right? Check the useful tips page, LasCon Storage - DFHSM Dataset Level Commands, at the link below: http://www.lascon.co.uk/d015401.htm Scott Barry SBBWorks, Inc. -- 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
Re: Need MASS recall advice
On Fri, 21 Mar 2008 10:25:01 -0500, Rugen, Len [EMAIL PROTECTED] wrote: Is there a faster way? If I had a list of files they wanted, could I send the recalls in tape - file sequence somehow to batch them? AFAIK , you do not even need to do explicit Hrecall If you have the list , you just run a BR14 and it will trigger an implicit hrecall . The list sorted by volume , in order to reduce the number of mount should improve performance . Bruno Bruno(dot)sugliani(at)groupemornay(dot)asso(dot)fr -- 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