David, Many moons ago (maybe DFSORT 1.7) the recommended total SORTWK space was 1.7 times the input file size. I've noted how using Hiperspace and Dataspace reduces the sortwk space used, and I'm wondering if your 10% recommendation is based on Sort being able to use an optimal amount of memory.
If so, would you recommend a larger value for SORTWK space on memory constrained system, or where customers have hobbled sort's memory in the ways that are often discussed on this list? In other words, is 110% good enough if sort chooses the smallest dynamic memory options. Some questions on your SORTWK performance tests: Does CFW on/off affect the impact when many SORTWK datasets are allocated (I'm assuming you mean 16 or more), and does the difference between 4 and many SORTWK change in any way when there is an intermediate merge? And a query: is the recommendation of 4 SORTWK good up to a particular size. I'm interested in that point because four SORTWK would be limited to about 17K CYLS total. Ron > -----Original Message----- > From: IBM Mainframe Discussion List [mailto:ibm-m...@bama.ua.edu] On Behalf Of > David Betten > Sent: Friday, February 12, 2010 6:39 AM > To: IBM-MAIN@bama.ua.edu > Subject: Re: [IBM-MAIN] SORTWK files > > For DFSORT, we generally recommend 4 work data sets. I have done testing > with varied numbers of work data sets and not seen much variance in > performance until you start allocating very large numbers of work data sets > for small sorts. Also, I would recommend that you allocate total space > of about 10% more than the file size. So for your 100 cylinder data set, > you'd want to allocate 4 work data sets of about 28 cylinders each. > > Also, we generally don't recommend allocating your own work data sets and > instead let DFSORT's dynamic allocation handle it. DFSORT can take into > account how much central storage is going to be used and adjust the work > data set space allocations accordingly. > > > Have a nice day, > Dave Betten > DFSORT Development, Performance Lead > IBM Corporation > email: bet...@us.ibm.com > DFSORT/MVSontheweb at http://www.ibm.com/storage/dfsort/ > > IBM Mainframe Discussion List <IBM-MAIN@bama.ua.edu> wrote on 02/12/2010 > 09:13:53 AM: > > > [image removed] > > > > SORTWK files > > > > Larry Crilley > > > > to: > > > > IBM-MAIN > > > > 02/12/2010 09:15 AM > > > > Sent by: > > > > IBM Mainframe Discussion List <IBM-MAIN@bama.ua.edu> > > > > Please respond to IBM Mainframe Discussion List. > > > > I want to dynamically allocate my own sort work files. If I know how > large > > my input file is (I can read the catalog), then what would everyone > > recommend for the allocation of the sort work files? How many datasets > are > > optimal? If my input file is 100 cylinders, for example, would you > allocate > > 3 sort work files CYL(35,0), 4 sort work files CYL(25,0), 8 sort work > > files.? > > > > > > > > > > > > Larry Crilley > > > > Dino-Software Corporation > > > > 800.480.DINO > > > > 412.366.3566 > > > > > > > <outbind://92-000000000C19A532C55CD94285A5E250EF9EC5A7445E7B00/www.dino-soft > > > ware.com> www.dino-software.com > > > > > > > > <blocked::blocked::http://www.dino-software.com/trex_factsheet.php> > T-REX - > > Superior ICF catalog mgmt with full Tape support and HSM Auditing > > <blocked::blocked::http://www.dino-software.com/reorgadon_factsheet.php> > > REORGadon - First ever online REORG for HSM > > <blocked::blocked::http://www.dino-software.com/teradon_factsheet.php> > > TERADON - First ever REPRO MERGECAT While-OPEN > > <blocked::blocked::http://www.dino-software.com> XTINCT - Secure > DASD/TAPE > > data eradication > > <blocked::blocked::http://www.dino-software.com/rtd_factsheet.php> RTD - > > DASD Real Time Defrag > > <blocked::blocked::http://www.dino-software.com/dal_details.php> DAL - > DINO > > healthcheck Analysis service for Legato > > <blocked::blocked::http://www.dino-software.com/sentinel_factsheet.php> > > SENTINEL - Real-time FTP Management. All Secure, all the time. > > > > > > ---------------------------------------------------------------------- > > For IBM-MAIN subscribe / signoff / archive access instructions, > > send email to lists...@bama.ua.edu 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 lists...@bama.ua.edu 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 lists...@bama.ua.edu with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
