on 10-21-2012 3:24 PM
Hello everyone,
We are facing with some some performance issues in our BW production system. We don´t know what´s the cause for this, so we request some help of you to support us in way to find the solution! We have the SAP BPC installed (SAP BusinessObjects Planning and Consolidation), the BPC when it do a call to SAP BW is running a chain of processes called ZBPC_AUTOMATISMOS_RH (transaction RSPC), this process chain makes a call to the class method in ZCL_HR_IGIF_UJ_CUSTOM_LOGIC~EXECUTE.
For a particular object that runs with a small filter data, between 8.000and 20.000 records, for only a cost center, it performs quickly, nearly 1 minute, but with a greater filter of data it takes much longer (isn´t proporcional!), with 200.000 records it takes something like 40 minutes!
Let me put here the values of some parameters that I changed after system installation (the server has 64Gb and 97Gb of Paging file):
- SAP Parameters adjusted:
PHYS_MEMSIZE 39000
abap/buffersize 2000000
rsdb/ntab/entrycount 50000
rsdb/ntab/ftabsize 100000
rsdb/ntab/sntabsize 10000
rsdb/ntab/irbdsize 15000
rsdb/cua/buffersize 15000
zcsa/presentation_buffer_area 15000000
zcsa/table_buffer_area 120000000
zcsa/db_max_buftab 7000
rtbb/buffer_length 30000
rdisp/max_wprun_time 3600
gw/cpic_timeout 60
gw/max_conn 2000
gw/max_overflow_size 25000000
rdisp/max_comm_entries 2000
icm/host_name_full ren-ed-43.ren.pt
icm/keep_alive_timeout 3600
rdisp/plugin_auto_logout 3600
em/global_area_MB 768
- Oracle Parameters adjusted:
alter system set sga_max_size=19G scope=spfile;
alter system set db_cache_size=13G scope=spfile;
alter system set shared_pool_size=3G scope=spfile;
alter system set pga_aggregate_target=3G scope=spfile;
This parameters adjustments are correct?
Can you help us please!?
Kind regards,
samid
Dear Samir,
As I can see there is a Z program involved kindly ask the author of this program to check Please inform the author and check with trace SE30.
Thanks
Tabrayz
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hello,
The server has 24 CPU´s, but when running a chain of processes as so far we have verified that only one CPU seems to be utilized and the CPU is running above the optimal maximum limit of 65 - 70%.
I think we can find further information on memory usage and paging rates and verify if the system configuration is optimal in those areas. Can you help me please how can I check the following information:
# Paging rate is over 20% of RAM per hour?
# Is the files system cache > 10% of RAM?
# What is the memory consumption of the program invoked when the logic
is running?
Can you help me please?!
Thank you,
samid
Hello Samid,
Before we jump into server parameters have you traced this Process Chain while its running ?
While monitoring did you see any behavior in your application server that made you decide to tweak the parameters ?
From what you have mentioned above it seems smaller filter works fine, wide range taking time. One approach would be
Chances are high you would be able to identify the root cause.
Thanks,
Samik
Hello Samik,
Ok... I tested it one more time and once again I checked some things, during the processing, I saw through SM50 a background process and in here, during the time running, I saw several Z reports passing, but none of them was actually stuck, and I never saw in there a SQL Statement, for the analysis with more detail in ST05, but unfortunately I never saw any SQL Statement in that background process!
During the processing I also tried to trace in ST05 ("trace with filter" option) the actions of user BPC_SYSADMIN (which is the user that ran the own process) but without any results, I don´t know why...
As I said in my previous message, this process only use one or two CPU´s of 24 CPU´s that exist in this server... and I don´t know why it no longer uses more CPU´s!! Why?!
Next I will post here some print-screens that I took during the process!
Kind regards,
samid
Hi Samid,
Firstly, ST05 should capture trace details while you set it with that user. Did you set the trace in the same application server where the job/processes are running ?
One more thing - try these, you should get some data
Usually parallel processing consuming multiple CPU for a particular job depends on application logic, not a basis task. And it is quite complex, mostly they do serial processing.
However, we should again try to trace and find the root cause.
Thanks,
Samik
What is the value for following:
abap/heap_area_total
abap/heap_area_nondia
abap/heap_area_dia
and Oracle parameter sga_target .
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
User | Count |
---|---|
82 | |
10 | |
10 | |
9 | |
6 | |
6 | |
5 | |
5 | |
4 | |
3 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.