home *** CD-ROM | disk | FTP | other *** search
- Comments: Gated by NETNEWS@AUVM.AMERICAN.EDU
- Path: sparky!uunet!zaphod.mps.ohio-state.edu!darwin.sura.net!paladin.american.edu!auvm!VCCSCENT.BITNET!SOMITCW
- Message-ID: <IBM-MAIN%92073115003305@RICEVM1.RICE.EDU>
- Newsgroups: bit.listserv.ibm-main
- Date: Fri, 31 Jul 1992 15:55:55 EST
- Sender: IBM Mainframe Discussion list <IBM-MAIN@RICEVM1.BITNET>
- From: SOMITCW@VCCSCENT.BITNET
- Subject: Re: UNIT and VOLUME specification
- Lines: 87
-
- On Fri, 31 Jul 1992 07:49:02 EDT, BENSON@POKVMCR3.VNET.IBM.COM said:
- >>Ah hah. So the real reason we don't get more user-friendly systems
- >>from IBM is that it's too hard to change the existing ancient cruddy
- >>code.
- >With our limited resources we have to prioritize the work and often
- >"pet peeves" are at the bottom. First, we fix our errors. Next, we
- >try to add new function, like SMS and tape libraries, to stay
- >competitive.
-
- What IBM considers "pet peeves", its customers sometimes consider
- broken or poorly designed code. For example:
-
- A few weeks ago a user submitted a job that looped passing records
- to SORT. SORT allocated all of the space on all work volumes causing
- other jobs to go into allocation recovery. All jobs were canceled
- each time the first user tried to run their job. Should I explain
- to all of the users that IBM broke allocation recovery 15 years ago
- or that IBM considers their problems as "pet peeves"?
-
- I have seen a 1 hour restore of a single VSAM cluster take over
- 5 hours while hundreds of users were waiting to use it in CICS.
- Should I explain to the users that the person setting up the job didn't
- understand that all VSAM clusters should always under all condition
- be defined specifying SPEED, but IBM defaults to a totally useless
- option of RECOVERY and that the default BUFNI, BUFND, and BUFSP must
- always be overridden.
-
- We have had to write over unknown user data sets because when
- IBM renamed IBCDMPRS ( Stand alone Dump/Restore ) to ADRDMPRS they
- decided to remove the DUMP function. I hate being the one to tell
- all users at 23 colleges that they may have lost a data set or two.
-
- We have been unable to copy and transmit data to other location
- because IBM and other vendors write reasonable block sizes but QSAM
- and BSAM only support less than half of 65535. Should I try to
- explain to users that a disk track is 47476 but they should write
- in half of a track blocks because QSAM and BSAM have decided to
- treat the half-word block size as signed. ( Note: Half of a track
- is not 23738 but is 23476. )
-
- IEBCOPY copies large blocks into data sets with smaller block sizes.
- IEBCOPY then destroys the directories of the output data set if a user
- tries to compress the data set the same day or months later. This has
- cost us and our users much grief.
-
- IEHMOVE is slow. It reloads its subroutines thousands of times
- in a run. It uses extremely inefficient work files.
- It doesn't support VIO for work files or for input or output files.
- It updates PDS directories between each member written.
- With the long run times and high disk contention it makes our users
- and operators unhappy. We don't run a data center planning to make
- our users unhappy.
-
- We are applying MVS maintenance now by SOURCEID. Could someone at
- IBM tell the engineer if PTFs with SOURCEIDs SMCCOR or SMCREC are
- from the CBPDO tape that came today or from 3 years ago?
- It's a bit late now, the engineer has loaded the entire CBPDO tape to
- disk and edited the ++ ASSIGN statements.
-
- Right now, we have users that need the online VSAM clusters and
- AIXes copied every night so they can use them the next morning in TSO
- and batch. DF/DSS does not allow us to rename the DATA and INDEX
- components during the copy and generates duplicate time stamp names
- for the clusters data and AIXes data component so the copy fails.
- The generated names also look like garbage in our LISTVTOCs and
- are impossible to code for in our security system rules.
- DF/DSS has been working on the problem for over 2 months. Of course
- they never did fix 2 previous problems we sent the a couple of years
- ago ( A backup of a single record VSAM data set can take 3 or more
- tapes for a single density 3380 ).
-
- Yesterday I called SyncSort Inc. with a suggestion that they include
- record counts in SYNCGENR. The person I spoke with said 'Thank You',
- she said that she would take the suggestion to the development manager,
- she said the suggestion would be added to the data base of suggestions.
- Could anyone imagine calling the IEBGENER support group and getting
- the same response? Would they use terms like "pet peeves" or claim
- limited resources to fix customer problems with software that we pay
- dearly for?
-
- If IBM is hearing that users need JCL improved so the users do not
- have to code a unit parameter for online ready disk, IBM should
- listen.
-
- I would like to make it clear that IBM has done a good job of fixing
- dozens of problems that I reported, but I also hope that the above
- list of problems and many more not listed will not exist next year.
-