View previous topic :: View next topic
|
Author |
Message |
seahawk789
New User
Joined: 22 Feb 2010 Posts: 56 Location: Cochin
|
|
|
|
Hi All,
I am exploring options to figure out the best way to identify region with maximum bandwidth to run jobs. Is there a way in Cobol program to get the % utilization of the LPAR ?
Thanks |
|
Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008 Posts: 8696 Location: Dubuque, Iowa, USA
|
|
|
|
It is not at all clear what you are asking. A batch job runs in an address space which will be given a certain amount of memory (the precise amount depends upon the site, the exits installed, and the parameters used at the site such as job class, which may have a memory limit). That address space has that amount of memory and it does not matter whether the COBOL program uses 1% or 100% of the memory (although using 100% could cause issues including ABENDS). When the program is not being executed, the system swaps out the address space and swaps in another one. Asking about % utilization of the LPAR makes no sense at all -- are you still talking memory, or are you talking about CPU usage? When you use the term "region", are you referring to CICS or batch jobs? And most sites allow different job classes to use different resources -- so if this is what you are asking, you MUST contact your site support group as we don't work at your site and have no idea what resources are allocated to each job class.
Bandwidth is a term used with networking -- it has nothing to do with memory and memory utilization.
Tell us what problem you are experiencing and we'll help if we can. But don't expect us to read your mind and figure out what you're asking. |
|
Back to top |
|
|
seahawk789
New User
Joined: 22 Feb 2010 Posts: 56 Location: Cochin
|
|
|
|
Robert Sample wrote: |
It is not at all clear what you are asking. A batch job runs in an address space which will be given a certain amount of memory (the precise amount depends upon the site, the exits installed, and the parameters used at the site such as job class, which may have a memory limit). That address space has that amount of memory and it does not matter whether the COBOL program uses 1% or 100% of the memory (although using 100% could cause issues including ABENDS). When the program is not being executed, the system swaps out the address space and swaps in another one. Asking about % utilization of the LPAR makes no sense at all -- are you still talking memory, or are you talking about CPU usage? When you use the term "region", are you referring to CICS or batch jobs? And most sites allow different job classes to use different resources -- so if this is what you are asking, you MUST contact your site support group as we don't work at your site and have no idea what resources are allocated to each job class.
Bandwidth is a term used with networking -- it has nothing to do with memory and memory utilization.
Tell us what problem you are experiencing and we'll help if we can. But don't expect us to read your mind and figure out what you're asking. |
Sorry for not being specific. Actually we have a series of batch jobs that connect to Online CICS modules through CICS Pipes. Sometimes when the CICS region is overloaded (High CPU usage) with large amount of online transactions, the jobs that try to connect to the CICS region is put on hold until CPU usage comes down for that region. Currently the CICS region to which the job connects is hardcoded.
We recently observed that there might be availability of CPU in other online regions, which the jobs could use. But as it is hardcoded we have no way to make the job run in that available region.
So I was looking to identify if there are any APIs/Functions that could be leveraged to efficiently determine the CICS regions with low CPU usage so that the job can connect to online programs and execute the program in that region.
Thanks |
|
Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008 Posts: 8696 Location: Dubuque, Iowa, USA
|
|
|
|
You have made a number of assumptions that you would need to verify with your site support group BEFORE going any further:
1. The different CICS regions each have access to the same programs. This is rarely true. Each region has its own DFHRPL, and thus programs may -- or may not -- be shared between regions.
2. Jobs connecting to the region are put on hold due to high CPU usage. This is rarely, if ever, true. Jobs won't execute because MAXTASKS is reached; as long as MAXTASKS isn't reached then the job can connect to a region with high CPU usage. There may be other constraints that impact your job execution (look up short-on-storage conditions in CICS for example).
3. High CPU usage may -- or may not -- be related to MAXTASKS; you could have a single transaction using lots of CPU time.
4. Since CICS regions are dynamic, a region with high CPU usage now may have almost no CPU usage 2 minutes from now (and vice versa).
5. How do you plan on determining whether or not a given CICS region has high CPU usage (or MAXTASKS or whatever your limiting factor is)? This is NOT a simple thing to determine. There is no API or function to provide this so you'll have to write code. |
|
Back to top |
|
|
|