on 01-12-2012 11:30 AM
hi,
sometimes our sap system become very slow and after some it will ok it occurs several times in a week.can anyone tell me how to resolve actual problem. i tried all tcode like sm50,st02,st06... but i dont know how to reconize actual problem and where sap go slow.
Hi Pawan,
In order to resolve the problem we need to understand what the problem is.Its a very generic statement that the system is slow.
You need to let us know as to which part of the system you seem to be having issues.
As suggested by others you need to fiirst segregate as to where the problem is os , db , sap.
If the issue reoccurs
OS : let us know how is the paging and memory and cpu in os07n
db : dbacockpit or st04old
then we can check for sap using stad as to where the issue might be or st03n
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
Hi Pawan,
I assume that what you said it is a system level performance issue not an individual transaction/site issue. When the issue is happThere are many possible causes which can result in a system level performance issue. so what is starting point to track this down? I think the key is to understand where an affected transaction/program spends time. Whether it is spent in dispatcher queue waiting for available work processes, significant roll-in/rollout( memory issue), long process time but little cpu time( could be CPU contention), long database time( database issue) etc...
SAP transaction STAD can show where the time spend. You can compare the statistical records of a common transaction executed both when system is normal and when system is slow. This normally helps to identify next steps.
Also when system is slow, check system has enough "free" work process, enough CPU on database server( utilization < 80%), health of database lock status (DB01), health of application lock status, enough CPU on application server, enough memory(extended memory and heap memory utilization <80%).
Swap in non program areas should be avoided but some of swap in those area should not have significant impact. When you deal with ST02 setting, combination of buffer size and number of entry makes difference. Make sure both have free space.
Finally, you might watch the situation whether there is a pattern.. it is often found that one particular application which is a culprit - for example, system slow can be caused by shortage of work processes. Shortage of work process can happen if one application is launching many parallel processes and use up majority or all available work processes such as dialog etc...
Hope this help
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
I would start with a simple top on the server, just to get an idea where i have to look further.
Cheers Michael
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
hi deepak,
thanks for giving me reply. in st02 there is always swap space shown for e.g.
Buffer Swaps
Nametab (NTAB) 0
Table definition 0
Field definition 3,118
Short NTAB 0
Initial records 6,944
program 0
CUA 4,438
Screen 0
Calendar 0
OTR 0
Tables 0
Generic Key 6
Single record 0
Export/import 0
Exp./ Imp. SHM 0
And the cpu and RAM utilization is shown as below
CPU
Utilization user % 3
system % 1
idle % 72
io wait % 24
System calls/s 8979
Interrupts/s 1966
Memory
Physical mem avail Kb 16755292 Physical mem free Kb 495704
now tell me what i do at that time.
Hi Pawan,
As per the information shared by you, first step is to tune your buffer.
These are the buffers which swap.
Field definition 3,118
Initial records 6,944
CUA 4,438
What you need to do ?
1) Check which are the profile parameters related to above affected buffers. You can double click the buffer entries and then click on current objects to get a list of profile parameters related to that buffer.
2) Adjust the identified parameters by increasing the values
Later monitor the impact.
Second step:
=========
What is the database version you use ?
Check whether database parameters are tuned as per SAP recommended notes.
Regards,
Deepak Kori
Hi,
Since how many days you are facing Performance Issue in your SAP system ?
Any activities are done at SAP System (Application Server) level ? Database Level ? Os Level ?
Performance Analysis, Tuning & optimization is ONGOING Process.
As suggested by Venkatesh, you should first identify the problems weather it is at SAP System (Application Server) level, or at Database level or at Operating System level. Its Joint efforts of SAP/DB/OS team. After Identification of the possible causes, you should have to start your analysis at the affected Level. After guided analysis you can find the root cause of the Performance degradation.
You will need to go through the following REFERENCE SAP Notes to compare existing Parameter values set in your SAP Landscape.
Note 425207 - SAP memory management, current parameter ranges
Note 941735 - SAP memory management for 64-bit Linux systems
Note 146289 - Parameter Recommendations for 64-Bit SAP Kernel
Note 1171650 - Automated Oracle DB parameter check
Anyway, What its the value you see in ST04 for "Log.Reads/User Calls" under section Calls. If this ratio is greater than 30 than it is strongly indicating that your SAP system is running with Expensive SQL statements.
Linux Kernel release level ?
Once you provide the required details, one can guide you for the possible reason/solution.
Regards,
Bhavik G. Shroff
Hi Bhavik,
I'm attempting to troubleshoot performance issues on our SAP central instance, virtualized, production, ECC6/db2/Linux server.
I don't see the "Read/User Calls" ratio referred to. Am I in the correct screen; Transaction DB6PER > Calls tab?
Sorry for some reason I couldn't upload an image.
Calls
=======================================================================================================
Read 477,410,972,291 SELECT SQL 1,333,698,405
Selected 6,126,526,213 UPDATE/INSERT/DELETE 269,744,667
Deleted 18,364,021 DDL 2,124
Inserted 32,385,658 Elapsed Time (sec) 2,375,017
Updated 759,473,624 Elapsed Time (microsec) 559,477
Rows Read / Rows Proc. 68.8
BP Gets / Rows Proc. 26.4
Total 2,590,740 COMMITs 192,432,754
Total Hash Loops 9,387 Rollbacks 351,351
Overflows 7,764 Dynamic SQL 2,368,065,972
Small Overflows 1,328 Static SQL 194,689,407
Post Threshold 0 Failed SQL 6,929,697
This ratio would definitely provide me with some insight.
Regards,
Mike
Hi Pawan,
Identify the bottlenecks in the system. Perf tuning is a vast area.
Start Analyzing from OS, DB and SAP.
Below are some basic aspects to look at.
OS
RAM utilization
CPU utilization
Swap utilization
DB
Instance parameters
Buffer Quality
Utilization of Buffer
SAP
ST03 Performance analysis
ST02 buffer utilization, Heap and extended memory areas
AL08 No. of users at different times
ST07
the list goes on, but the above should help you the initial analysis.
Br,
Venky
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.