Re: A deep question about VSAM SHR(4) - can you help?
Um, be careful about drawing any conclusions from a simple test other than the syntax is close. Data sharing is a very complex issue with gotcha's aplenty. You cannot possibly test every variation of access, update, and timings. The VSAM folks have, IMHO, done a fine job of preventing you from screwing up too bad. But VSAM is still not a DBMS and doesn't claim to be. If the philosophy cannot change, then maybe neither can the solution. Yes, it can be really that simple. Now, this is not to say that one couldn't forge right on and things would seem to work just fine, maybe for years. Or maybe not. 'Unpredictable results' include seemingly successful tests. Imagine corruption slowly creeping in and not discovered until the last good backup has rolled off and the tape reused. I believe what you want to do is doable. But you just gotta follow the rules. If you do, then CICS and VSAM can work together to avoid corruption and enhance performance. -Original Message- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@bama.ua.edu] On Behalf Of Mike Kovach Sent: Thursday, April 05, 2012 3:23 PM To: IBM-MAIN@bama.ua.edu Subject: Re: A deep question about VSAM SHR(4) - can you help? Thanks Robert. That is exactly what I was thinking. Will be running a test early next week. Regards, Mike Kovach From: Robert A. Rosenberg hal9...@panix.com To: IBM-MAIN@bama.ua.edu Sent: Thursday, April 5, 2012 4:03 PM Subject: Re: A deep question about VSAM SHR(4) - can you help? At 07:16 -0700 on 04/05/2012, Mike Kovach wrote about A deep question about VSAM SHR(4) - can you help?: I have a VSAM KSDS CLUSTER which is written to by ONLY ONE PROGRAM in ONLY ONE CICS REGION. Currently, this file is defined in CICS with STRNO(1). The file is defined with SHR(4,3) because while being written ONLY in CICS, it is being read by a non-reentrent ASSEMBLER program running in BATCH. SHR 4 forces VSAM to harden each I/O (yeah, I know!) so the BATCH gets the current information. Please spare me all the comments about how poor this solution is as it stands. It has been in place for decades and due to a myriad of reasons, the philosophy CANNOT change. My specific question is this: I want to introduce multi tasking so that 5 copies of the program can update the file concurrently. If we change STRNO(1) to STRNO(5) on the CICS FCT Definition, will VSAM be smart enough to manage the writes to the file so we don't break it and the BATCH still gets the current information? So long as you are still using one CICS Region, I do not think you will run into problems. The STRNO(5) will allow you to have 5 CIs being updated at a time (one CI per copy of the program). If more than one copy attempts to access records from the same CI, it should cause the subsequent requesters to wait for the owning copy to finish its update and release/write the CI (just make sure that all your VSAM is being done by SubTasks which I think CICS does automatically). You should increase the number of buffers so there are enough for all the copies. I am interested in any discussion you might share, but I am most interested in a specific reference to a reliable document. Please help. Thanks Mike Kovach -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN NOTICE: This electronic mail message and any files transmitted with it are intended exclusively for the individual or entity to which it is addressed. The message, together with any attachment, may contain confidential and/or privileged information. Any unauthorized review, use, printing, saving, copying, disclosure or distribution is strictly prohibited. If you have received this message in error, please immediately advise the sender by reply email and delete all copies. -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN
Re: A deep question about VSAM SHR(4) - can you help?
CICS moderates access to a VSAM file by doing internal CICS-level ENQs. For a KSDS, the ENQ is on the key value. A VSAM dataset in a single CICS region can be used and updated safely my multiple transactions running in that region (or in multiple regions if the file request is sent to a file owning region). Of course, this can lead to a dead lock situation if the transactions try to update multiple records in the same dataset, unless something is done to exclude it. If the VSAM file is local to multiple regions, or between CICS and batch, then you can corrupt the data. If you need concurrent access via batch and CICS, I suggest HW's SYSB product. We use it and it works with no real problems. Basically, SYSB extends CICS' support for Multi Region Option (MRO) access to a dataset to batch. The actual I/O is done in the CICS region. Another post mentioned that two records in the same CI could not undergo update processing at the same time. This is true. However, the current CICS versions use VSAM options such that VSAM returns a indicator on the second request to CICS. And CICS moderates this. http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/dfhp3c00/4.2.6.1 quote If you do not use VSAM record-level sharing, CICS serializes update requests by base cluster record key. The complete VSAM control interval (CI) containing the requested record is held for exclusive use while an individual command (for example, a READ command with the UPDATE option) is being executed on the record. Once each command is complete, the control interval is released, and only the requested record remains locked. For nonrecoverable data sets, both the VSAM exclusive use and the CICS exclusive use of the record end when the update request is complete in VSAM terms; for example, when the REWRITE command has completed. For recoverable data sets, however, the CICS exclusive use does not end until the task ends or issues a SYNCPOINT command. Recoverability is specified in the data set resource definition. See FILE definition attributes in the CICS Resource Definition Guide for more information about the FILE resource definitions. /quote -- John McKown Systems Engineer IV IT Administrative Services Group HealthMarkets® 9151 Boulevard 26 . N. Richland Hills . TX 76010 (817) 255-3225 phone . john.mck...@healthmarkets.com . www.HealthMarkets.com Confidentiality Notice: This e-mail message may contain confidential or proprietary information. If you are not the intended recipient, please contact the sender by reply e-mail and destroy all copies of the original message. HealthMarkets® is the brand name for products underwritten and issued by the insurance subsidiaries of HealthMarkets, Inc. -The Chesapeake Life Insurance Company®, Mid-West National Life Insurance Company of TennesseeSM and The MEGA Life and Health Insurance Company.SM -Original Message- From: IBM Mainframe Discussion List [mailto:IBM-MAIN@bama.ua.edu] On Behalf Of Walt Farrell Sent: Thursday, April 05, 2012 6:16 PM To: IBM-MAIN@bama.ua.edu Subject: Re: A deep question about VSAM SHR(4) - can you help? On Thu, 5 Apr 2012 07:16:15 -0700, Mike Kovach mrmach...@yahoo.com wrote: My specific question is this: I want to introduce multi tasking so that 5 copies of the program can update the file concurrently. If we change STRNO(1) to STRNO(5) on the CICS FCT Definition, will VSAM be smart enough to manage the writes to the file so we don't break it and the BATCH still gets the current information? I am not a VSAM expert, nor a CICS expert (nor am I sure whether your program is using CICS functions to write to the data set, or using VSAM macros directly), but I would be concerned about serialization. From DFSMS Using Data Sets at http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/dgt 2d4a0/2.7.2.1?SHELF=EZ2ZO213DT=20110606092005 or http://preview.tinyurl.com/76joxao you can read that for SHAREOPTIONS 4 you have the same serialization requirements as for SHAREOPTIONS 3, and for SHAREOPTIONS 3 the book says quote This option requires that the user's program use ENQ/DEQ to maintain data integrity while sharing the data set, including the OPEN and CLOSE processing. User programs that ignore the write integrity guidelines can cause VSAM program checks, lost or inaccessible records, uncorrectable data set failures, and other unpredictable results. This option places responsibility on each user sharing the data set. /quote So unless there's something in CICS issuing appropriate ENQ/DEQ macros, I think you'll need to make some program changes. -- Walt -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe
Re: A deep question about VSAM SHR(4) - can you help?
At 07:16 -0700 on 04/05/2012, Mike Kovach wrote about A deep question about VSAM SHR(4) - can you help?: I have a VSAM KSDS CLUSTER which is written to by ONLY ONE PROGRAM in ONLY ONE CICS REGION. Currently, this file is defined in CICS with STRNO(1). The file is defined with SHR(4,3) because while being written ONLY in CICS, it is being read by a non-reentrent ASSEMBLER program running in BATCH. SHR 4 forces VSAM to harden each I/O (yeah, I know!) so the BATCH gets the current information. Please spare me all the comments about how poor this solution is as it stands. It has been in place for decades and due to a myriad of reasons, the philosophy CANNOT change. My specific question is this: I want to introduce multi tasking so that 5 copies of the program can update the file concurrently. If we change STRNO(1) to STRNO(5) on the CICS FCT Definition, will VSAM be smart enough to manage the writes to the file so we don't break it and the BATCH still gets the current information? So long as you are still using one CICS Region, I do not think you will run into problems. The STRNO(5) will allow you to have 5 CIs being updated at a time (one CI per copy of the program). If more than one copy attempts to access records from the same CI, it should cause the subsequent requesters to wait for the owning copy to finish its update and release/write the CI (just make sure that all your VSAM is being done by SubTasks which I think CICS does automatically). You should increase the number of buffers so there are enough for all the copies. I am interested in any discussion you might share, but I am most interested in a specific reference to a reliable document. Please help. Thanks Mike Kovach -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN
Re: A deep question about VSAM SHR(4) - can you help?
Thanks Robert. That is exactly what I was thinking. Will be running a test early next week. Regards, Mike Kovach From: Robert A. Rosenberg hal9...@panix.com To: IBM-MAIN@bama.ua.edu Sent: Thursday, April 5, 2012 4:03 PM Subject: Re: A deep question about VSAM SHR(4) - can you help? At 07:16 -0700 on 04/05/2012, Mike Kovach wrote about A deep question about VSAM SHR(4) - can you help?: I have a VSAM KSDS CLUSTER which is written to by ONLY ONE PROGRAM in ONLY ONE CICS REGION. Currently, this file is defined in CICS with STRNO(1). The file is defined with SHR(4,3) because while being written ONLY in CICS, it is being read by a non-reentrent ASSEMBLER program running in BATCH. SHR 4 forces VSAM to harden each I/O (yeah, I know!) so the BATCH gets the current information. Please spare me all the comments about how poor this solution is as it stands. It has been in place for decades and due to a myriad of reasons, the philosophy CANNOT change. My specific question is this: I want to introduce multi tasking so that 5 copies of the program can update the file concurrently. If we change STRNO(1) to STRNO(5) on the CICS FCT Definition, will VSAM be smart enough to manage the writes to the file so we don't break it and the BATCH still gets the current information? So long as you are still using one CICS Region, I do not think you will run into problems. The STRNO(5) will allow you to have 5 CIs being updated at a time (one CI per copy of the program). If more than one copy attempts to access records from the same CI, it should cause the subsequent requesters to wait for the owning copy to finish its update and release/write the CI (just make sure that all your VSAM is being done by SubTasks which I think CICS does automatically). You should increase the number of buffers so there are enough for all the copies. I am interested in any discussion you might share, but I am most interested in a specific reference to a reliable document. Please help. Thanks Mike Kovach -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN
Re: A deep question about VSAM SHR(4) - can you help?
On Thu, 5 Apr 2012 07:16:15 -0700, Mike Kovach mrmach...@yahoo.com wrote: My specific question is this: I want to introduce multi tasking so that 5 copies of the program can update the file concurrently. If we change STRNO(1) to STRNO(5) on the CICS FCT Definition, will VSAM be smart enough to manage the writes to the file so we don't break it and the BATCH still gets the current information? I am not a VSAM expert, nor a CICS expert (nor am I sure whether your program is using CICS functions to write to the data set, or using VSAM macros directly), but I would be concerned about serialization. From DFSMS Using Data Sets at http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/dgt2d4a0/2.7.2.1?SHELF=EZ2ZO213DT=20110606092005 or http://preview.tinyurl.com/76joxao you can read that for SHAREOPTIONS 4 you have the same serialization requirements as for SHAREOPTIONS 3, and for SHAREOPTIONS 3 the book says quote This option requires that the user's program use ENQ/DEQ to maintain data integrity while sharing the data set, including the OPEN and CLOSE processing. User programs that ignore the write integrity guidelines can cause VSAM program checks, lost or inaccessible records, uncorrectable data set failures, and other unpredictable results. This option places responsibility on each user sharing the data set. /quote So unless there's something in CICS issuing appropriate ENQ/DEQ macros, I think you'll need to make some program changes. -- Walt -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN
Re: A deep question about VSAM SHR(4) - can you help?
Another question to ponder; Is the file NONLSR or LSR in CICS? Another question; Does the BATCH program run in the same LPAR as the CICS? always? Another question; Are you using BLSR in the batch program? or any other buffer tuning software? Another question; Does the BATCH program open the file for input only? Regards, Doug Sent from my iPhone On Apr 5, 2012, at 16:23, Mike Kovach mrmach...@yahoo.com wrote: Thanks Robert. That is exactly what I was thinking. Will be running a test early next week. Regards, Mike Kovach From: Robert A. Rosenberg hal9...@panix.com To: IBM-MAIN@bama.ua.edu Sent: Thursday, April 5, 2012 4:03 PM Subject: Re: A deep question about VSAM SHR(4) - can you help? At 07:16 -0700 on 04/05/2012, Mike Kovach wrote about A deep question about VSAM SHR(4) - can you help?: I have a VSAM KSDS CLUSTER which is written to by ONLY ONE PROGRAM in ONLY ONE CICS REGION. Currently, this file is defined in CICS with STRNO(1). The file is defined with SHR(4,3) because while being written ONLY in CICS, it is being read by a non-reentrent ASSEMBLER program running in BATCH. SHR 4 forces VSAM to harden each I/O (yeah, I know!) so the BATCH gets the current information. Please spare me all the comments about how poor this solution is as it stands. It has been in place for decades and due to a myriad of reasons, the philosophy CANNOT change. My specific question is this: I want to introduce multi tasking so that 5 copies of the program can update the file concurrently. If we change STRNO(1) to STRNO(5) on the CICS FCT Definition, will VSAM be smart enough to manage the writes to the file so we don't break it and the BATCH still gets the current information? So long as you are still using one CICS Region, I do not think you will run into problems. The STRNO(5) will allow you to have 5 CIs being updated at a time (one CI per copy of the program). If more than one copy attempts to access records from the same CI, it should cause the subsequent requesters to wait for the owning copy to finish its update and release/write the CI (just make sure that all your VSAM is being done by SubTasks which I think CICS does automatically). You should increase the number of buffers so there are enough for all the copies. I am interested in any discussion you might share, but I am most interested in a specific reference to a reliable document. Please help. Thanks Mike Kovach -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN
Re: A deep question about VSAM SHR(4) - can you help?
At 18:16 -0500 on 04/05/2012, Walt Farrell wrote about Re: A deep question about VSAM SHR(4) - can you help?: This option requires that the user's program use ENQ/DEQ to maintain data integrity while sharing the data set, including the OPEN and CLOSE processing. User programs that ignore the write integrity guidelines can cause VSAM program checks, lost or inaccessible records, uncorrectable data set failures, and other unpredictable results. This option places responsibility on each user sharing the data set. /quote So unless there's something in CICS issuing appropriate ENQ/DEQ macros, I think you'll need to make some program changes. You are misreading the restrictions. This has to do with sharing the VSAM Cluster between multiple programs NOT sharing the Cluster between multiple tasks of the same program (as would be the case with CICS). You are using ONE ACB so there is no interlock issue. I am not sure what QNAME/RNAME is supposed to be used for the interlock. VSAM itself does the ENQ/DEQ (I think) via using the CI number (it can not be the key of the record because while that would prevent multiple updates to that record, you can mess up the CI if there are two records in the CI being updated) as part of the RNAME (it has been a while). -- For IBM-MAIN subscribe / signoff / archive access instructions, send email to lists...@bama.ua.edu with the message: INFO IBM-MAIN
