on 07-22-2015 9:36 AM
Hi,
For two months the system is very slow, we tried to make a performance
and we have seen that the number of buffers was too something.
To try to solve the problem, we perform the following steps:
1- scheduled a stop restart services and the physical server to clean semaphores, burffers, traces, logs, etc.
2- Analyze the memory parameters recommended by SAP (note 88416) are correct.
3- Finally update the Kernel version of the 7.20 to the 7.21 patch 500.
With all this, users still I complaining that the system do very slow, I put an example. "Simple balance inquiries with only two users working. Another user tells me that released a report (zinver1) one half and it took him five hours".
Thanks
Regards
Jose, in which kind of process do you see this slowness? Can you attach the screen of SMLG response time of the system?
There are a lot of variants that can impact SAP performance. First of all you need to identify if is all system is slow or just some processes.
In the most cases DB is the bad guy.
Are yours database jobs running(I mean house keeping like reorg, update,etc...)? What is the size of your databse? Is this a DEV system?
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Jose,
Has there been any changes into your system landscape, for ex any recent upgrades. Please check the following parameters,since slowness in a system can be due to varied reasons.
1. Confirm if your system is ABAP or Dual-Stack.
2. Is the slowness occurring at any particular period of the day, or the entire day.
3. The attached screenshot for ST02 shows huge swaps in CUA buffer. Could you please check that, and confirm if the swaps are occurring on a daily basis.
4. Are there some jobs which are consuming high resources.
5. Check the performance from ST03, CPU Utilization from ST06,
Also confirm the DB and OS types.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Sinha,
I answer to your questions.
1. Confirm if your system is ABAP or Dual-Stack.
My system is ABAP
2. Is the slowness occurring at any particular period of the day, or the entire day.
This goes happening constant for two months.
3. The attached screenshot for ST02 shows huge swaps in CUA buffer. Could you please check that, and confirm if the swaps are occurring on a daily basis.
We restart the system and update the version of the Kernel and the system clean the buffers.
4. Are there some jobs which are consuming high resources.
These jobs always have been this way and previously the system was not going slowly.
5. Check the performance from ST03, CPU Utilization from ST06,
The performance already I have verified it and the times are correct.
Hi Jose,
Did you check if there is any unexpected database growth. If i suppose the DB you are using is Oracle, could you please check if the tablespaces are having sufficient free space. Also, are the DB cleanup jobs running fine in DB13.
Also please check if some huge logs are getting generated in the system, which might lead to slowness. Also check if the load balancing is working fine in SMLG, and there are sufficient work-process in accordance with the number of CPUs.
Hi Jose Luis Gil,
1. the ST02 shows all buffers have hitrate of > 99%. There is no need for any action in this section!
2. the ST03 shows a high DB time. This indicates that the database is the problem.
Unfortunately I'm only familiar with ORACLE and ST04 show this is not an ORACLE DB.
But my experience is that the DB-times are most often the problem!
I would examine the most expensive SQL-statements and raise the DB-cache! DB-cache is the more important performance mechanism, because using the storage is most expensive in performance.
Best regards
Willi Eimler
Hi Jose Luis Gil,
could you please post a screenshot of transactions:
SM50 (please push the cpu button)
ST03 (please select total and last week).
ST02
Best regards
Willi Eimler
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Jose Luis Gil,
it seems that the SICF is not configured properly.
Please check if instance parameter:
icm/server_port_0 PROT=HTTP,PORT=8000
Than check transaction SICF.
Please activate:
Best regards
Willi Eimler
User | Count |
---|---|
88 | |
10 | |
10 | |
9 | |
7 | |
7 | |
6 | |
5 | |
4 | |
4 |
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.