View previous topic :: View next topic
|
Author |
Message |
nethaji_mainframer
New User
Joined: 12 Dec 2011 Posts: 3 Location: India
|
|
|
|
Hi,
We are using a Z800 Mainframe with 8GB of memory. Compilation jobs are comsuming more time. Could any one please give me some of the resons for slow compilation? |
|
Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
|
|
|
|
Has anyone been putting any extra code in the programs before compiling them?
Slower than what? Different machine? New version of z/OS? Any one of lots of other things which I'm only going to pretend I'm thinking about to try to keep you happy?
Get specific if you hope to get a more useful answer. |
|
Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008 Posts: 8696 Location: Dubuque, Iowa, USA
|
|
|
|
Quote: |
Could any one please give me some of the resons for slow compilation? |
1. System contention for CPU
2. System contention for disk
3. Larger programs being compiled.
4. Change in WLM policy
5. Change in JES
6. Change in z/OS
7. Using different disk drives
8. User error (submitting compile in worng class, for example)
and the list goes on and on....
Your post is chock full of fuzzy terms -- slow is not a good term since it is not clear if you mean the elapsed time of the compile job is longer, or the amount of CPU time being used is more, or the I/O count is hgiher, or even what you are comparing against to determine "slow" -- the same compile a month ago? a year ago? 2 days ago? a different job? compile jobs against production workload? |
|
Back to top |
|
|
nethaji_mainframer
New User
Joined: 12 Dec 2011 Posts: 3 Location: India
|
|
|
|
Our mainframe is used only for Training perpose. Only new user are going to use it. The codes are comperatively smaller less than 200 to 300 lines.
Every step in the Compilation takes more than 2 to 3 mins even after the initiators Picked the job for execution.
At present there are 700 to 800 user are using. Ther are 40 initiators are active.
Only compliation jobs are slow but execution jobs are much faster not more than 10 mill seconds, Once it is in execution. |
|
Back to top |
|
|
Binop B
Active User
Joined: 18 Jun 2009 Posts: 407 Location: Nashville, TN
|
|
|
|
Hi Nethaji,
Try executing the compilation job with the CLASS that is used for the exeuction of the program - in case they are not same. |
|
Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
|
|
|
|
But are you saying this is different from before, or you have complaints from the trainees that compiles take a lot of time?
If your machine is training-only, then someone should decide priorities/classes which suit the training being provided. Then the results of that should be implemented. |
|
Back to top |
|
|
Robert Sample
Global Moderator
Joined: 06 Jun 2008 Posts: 8696 Location: Dubuque, Iowa, USA
|
|
|
|
Quote: |
Every step in the Compilation takes more than 2 to 3 mins even after the initiators Picked the job for execution. |
Why is this seen as a problem -- who is setting the expectations? My site has a z800 with 8 GB of memory, and compiles are done in the same LPAR as production work. 5 to 10 minutes is not an unusual time for a compile, and 2 to 3 hours is not unheard of when the system is heavily loaded (such as when month-end production is running). |
|
Back to top |
|
|
dick scherrer
Moderator Emeritus
Joined: 23 Nov 2006 Posts: 19244 Location: Inside the Matrix
|
|
|
|
Hello,
Quote: |
Our mainframe is used only for Training perpose. Only new user are going to use it. |
Sounds like rather than managing this test system, everything is piled together in the same set of libraries - which intereferes with concurrent compiles/links.
Just as an expirement, define a new set of libraries for some group of students and the performance of those separated compiles should improve. |
|
Back to top |
|
|
Bill Woodger
Moderator Emeritus
Joined: 09 Mar 2011 Posts: 7309 Location: Inside the Matrix
|
|
|
|
Maybe you have a smart one, who's worked out how to do foreground compiles. Thirty-nine oher students are copying them and are so impressed with the turnaround that they are recompiling after each line added/changed.
I hope this is not a "reverse-psychology" marketing ploy. |
|
Back to top |
|
|
Pedro
Global Moderator
Joined: 01 Sep 2006 Posts: 2547 Location: Silicon Valley
|
|
|
|
Check for data set enqueues.
Perhaps the RMF Workload Delay Manager has been implemented... use
and explore all of the reports under RESOURCE.
If RMFWDM is not implemented, ask your sysprog to implement it. |
|
Back to top |
|
|
|