Re: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
In [EMAIL PROTECTED], on 04/02/2008 at 04:35 AM, Sascha Weng [EMAIL PROTECTED] said: I have a programm written in Cobol which runs monthly on a z/Os-System. The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles. It consumes nearly the same numbers of input records every month. I wonder now, why there are so great differences in the consumed cpu-time. First, you should expect significant variations in SRB time just from differences in system load. Second, the number of records may not be a reliable guide to how much work your job has to do. Is the average record length subject to variation from run to run? Are there differences in compression or encryption requirements? Is there processing that is only done in specific calendar periods? Has the program been changed to accommodate regulatory changes? -- Shmuel (Seymour J.) Metz, SysProg and JOAT ISO position; see http://patriot.net/~shmuel/resume/brief.html We don't care. We don't have to care, we're Congress. (S877: The Shut up and Eat Your spam act of 2003) -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
the job protocol 11.25 (what ever that means) and this month it consumed 16.33. Actually I expected the cpu time would be nearly the same. But there is this difference I can not explain. Any suggestions would be greatly appreciated. There are several reasons why Run- or CPU-Times may change. One reason could be that a extent has to be allocated (for one of your output datasets) when running the job is running (and another time this is not neccessary). Or depending of the values you insert into the DB2-Tables the INSERTs need more CPU/Runtime because the index(es) of your tables need to be reorganized or something like this. Bye, Michael -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
Differences in runtime can be a variety of reasons. Do you have any performance reporting tools like STROBE or do you know how to use RMF or SMF reporting functions? Do you have MICS or MXG? More or less, it could be due to any of the following 1) Changes in file size 2) Changes in workload during this run 3) Changes in DB2 workload 4) Other performance issues in the system at the time. Tape or DASD used by this job can impact run time. 5) How busy was your system this time compared to last time it ran? The numbers 11.25 means 11 mins and 25 seconds. That in itself is not always an issue. Did you meet your SLAs for this job? Jobs can vary in runtime. You need to have a baseline that says this job will usually run in 10 mins. If it runs over 20 mins it is a problem. Lizette -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Sascha Weng Sent: Wednesday, April 02, 2008 5:36 AM To: IBM-MAIN@BAMA.UA.EDU Subject: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication? It is the first time, that I write to this list and I am not quiet sure, if I am on the right place for asking my question. I hope not being off-topic. If so, please accept my apology for my ignorance in advance. I have a programm written in Cobol which runs monthly on a z/Os-System. The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles. It consumes nearly the same numbers of input records every month. I wonder now, why there are so great differences in the consumed cpu-time. For example: Last month the programm consumed due to the job protocol 11.25 (what ever that means) and this month it consumed 16.33. Actually I expected the cpu time would be nearly the same. But there is this difference I can not explain. Any suggestions would be greatly appreciated. -- 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
CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
It is the first time, that I write to this list and I am not quiet sure, if I am on the right place for asking my question. I hope not being off-topic. If so, please accept my apology for my ignorance in advance. I have a programm written in Cobol which runs monthly on a z/Os-System. The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles. It consumes nearly the same numbers of input records every month. I wonder now, why there are so great differences in the consumed cpu-time. For example: Last month the programm consumed due to the job protocol 11.25 (what ever that means) and this month it consumed 16.33. Actually I expected the cpu time would be nearly the same. But there is this difference I can not explain. Any suggestions would be greatly appreciated. -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
I forgot to mention that you should search the IBM MAIN archives, I believe there was a recent thread on job runtime variations. Lizette snip It is the first time, that I write to this list and I am not quiet sure, if I am on the right place for asking my question. I hope not being off-topic. If so, please accept my apology for my ignorance in advance. I have a programm written in Cobol which runs monthly on a z/Os-System. The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles. It consumes nearly the same numbers of input records every month. I wonder now, why there are so great differences in the consumed cpu-time. For example: Last month the programm consumed due to the job protocol 11.25 (what ever that means) and this month it consumed 16.33. Actually I expected the cpu time would be nearly the same. But there is this difference I can not explain. Any suggestions would be greatly appreciated. unsnip -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
On Wed, 2008-04-02 at 04:35 -0500, Sascha Weng wrote: The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles [...] I expected the cpu time would be nearly the same. But there is this difference I can not explain. A month or two ago we enabled compression for a handful of big sequential datasets. This significantly increased CPU consumption for readers, yet reduced the actual run time by half. Perhaps your storage admin gremlins have done something similar? -- David Andrews A. Duda and Sons, Inc. [EMAIL PROTECTED] -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
I have a programm written in Cobol which runs monthly on a z/Os-System. The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles. It consumes nearly the same numbers of input records every month. I wonder now, why there are so great differences in the consumed cpu-time. For example: Last month the programm consumed due to the job protocol 11.25 (what ever that means) and this month it consumed 16.33 If the COBOL program simply looped it would use a certain amount of CPU time and that amount of time would be more or less the same every time you ran it on a given machine. However your COBOL program does useful work, reading and writing data. It consumes system services while doing that and those system services are related to the amount of data being processed. So if the amount of data being processed is about the same then you would expect the amount of system resources (including CPU) to be about the same. You are seeing roughly a 5 second (45%) increase in CPU consumption, so that is far outside of what you might expect from random variations. Are you sure your program is processing the same amount of data? If the amount of data really is about the same then you have to look to the organization of the data. For your input files; the block size may have changed which could lead to an increase in CPU time, but it is unlikely that you would see such a big change. For your output; there are lots of minor changes that can have major impacts on performance in DB2. That pretty much leaves DB2 as the potential culprit. I would look there first. CC -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
And is it running on the same physical machine every time? Or are there multiple systems of different speeds where it can execute? Dave Thorn * Senior Technology Analyst * SunGard Computer Services * 600 Laurel Oak Road, Voorhees, NJ, 08043 Office 856 566-5412 * Mobile 609 781-0353 * Fax 856 566-3656 CONFIDENTIALITY: This e-mail (including any attachments) may contain confidential, proprietary and privileged information, and unauthorized disclosure or use is prohibited. If you received this e-mail in error, please notify the sender and delete this e-mail from your system. -Original Message- From: IBM Mainframe Discussion List [mailto:[EMAIL PROTECTED] On Behalf Of Craddock, Chris Sent: Wednesday, April 02, 2008 10:22 AM To: IBM-MAIN@BAMA.UA.EDU Subject: Re: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication? I have a programm written in Cobol which runs monthly on a z/Os-System. The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles. It consumes nearly the same numbers of input records every month. I wonder now, why there are so great differences in the consumed cpu-time. For example: Last month the programm consumed due to the job protocol 11.25 (what ever that means) and this month it consumed 16.33 If the COBOL program simply looped it would use a certain amount of CPU time and that amount of time would be more or less the same every time you ran it on a given machine. However your COBOL program does useful work, reading and writing data. It consumes system services while doing that and those system services are related to the amount of data being processed. So if the amount of data being processed is about the same then you would expect the amount of system resources (including CPU) to be about the same. You are seeing roughly a 5 second (45%) increase in CPU consumption, so that is far outside of what you might expect from random variations. Are you sure your program is processing the same amount of data? If the amount of data really is about the same then you have to look to the organization of the data. For your input files; the block size may have changed which could lead to an increase in CPU time, but it is unlikely that you would see such a big change. For your output; there are lots of minor changes that can have major impacts on performance in DB2. That pretty much leaves DB2 as the potential culprit. I would look there first. CC -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
I'd suggest that you look into DB2 detailed accounting report which gives a very good account of where the CPU was spent while doing DB2 work. Compare the two reports if your site keeps this historical data otherwise make do with new report and see what can be improved. There are so many factors affecting DB2 CPU usage that a guess is more worthless than you would think. Mohammad On Wed, 2 Apr 2008 04:35:56 -0500, Sascha Weng [EMAIL PROTECTED] wrote: It is the first time, that I write to this list and I am not quiet sure, if I am on the right place for asking my question. I hope not being off-topic. If so, please accept my apology for my ignorance in advance. I have a programm written in Cobol which runs monthly on a z/Os-System. The programm reads inputrecords and writes them in a database (DB2) and in different outputfiles. It consumes nearly the same numbers of input records every month. I wonder now, why there are so great differences in the consumed cpu-time. For example: Last month the programm consumed due to the job protocol 11.25 (what ever that means) and this month it consumed 16.33. Actually I expected the cpu time would be nearly the same. But there is this difference I can not explain. Any suggestions would be greatly appreciated. -- 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: CPU-Time in z/Os-System. Is The Cpu Time In z/Os-systems A Meaningful Indication?
Chris has said (better than I could) what I was thinking. First stop should be DB2. Jon snip If the amount of data really is about the same then you have to look to the organization of the data. For your input files; the block size may have changed which could lead to an increase in CPU time, but it is unlikely that you would see such a big change. For your output; there are lots of minor changes that can have major impacts on performance in DB2. That pretty much leaves DB2 as the potential culprit. I would look there first. /snip -- 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