cancel
Showing results for 
Search instead for 
Did you mean: 

Performance issues

Former Member
0 Kudos

Hi all,

I need some pointers as to how I can narrow down users' complain of poor response time/performance whenever they run a transaction.

I understand that there are many considerations e.g. PC specs/speed, network, etc that will affect users' performance but at least as far as the SAP servers are concerned, where the bottlenecks/choke points are.

From ST06, the paging is good as most of the hour-to-hour stats showed 0 or below the 100 range. This should mean that there are enough physical memory and is not a memory issue, right?

Is this the only stats I should be looking at?

Would anyone be able to advise on what are the steps that I should follow at least to say that the performance issue is not at the SAP end?

Thanks

Accepted Solutions (1)

Accepted Solutions (1)

Sriram2009
Active Contributor
0 Kudos

Hi ,

1.You can use the T code STAD type the user name and transaction code which he is using   you can check the result how much time taken in the process and database

2. During that time you can check the ST02 for memory tuning / st03 overall work load during the time

  3.  take the trace T code ST12. check the ABAP & DB time.

Regards

Sriram

former_member184548
Participant
0 Kudos

Hi

Whenever you get tickets for slowness of SAP system. Start from looking towards the basic tcodes for basis.

1. Narrow down the problem .. like how many users and what kind of users are facing the problem. Some times some set of users will face performance issue, it may be due to some changes which might have gone into production system recently. Look out for the latest imported transport request.

2. Space monitoring : db02 check space for all table space. Also check for PSAPUNDO

3. SAP Buffer Memory monitoring : ST02 : See the swapping in buffers

4. Look for the logs in SM21 and dump in st22

5. Check the status of your all work processes. Do you have sufficient work processes available. Check if you have proper number of update work processes available, Check the enqueue process, It should be in waiting status most of the time. If everything is fine then check the work process logs. Some times if you are logged into a application server and it may have some connection issue with the database instance. Check it with nipping.

6. CHeck the user distribution via log on groups

7. Check resource quota allocation for RFC group. You should not specify all work process to be used by RFC logon group.

8. Check for jobs which may be using parallel generator option which usually uses many dialog ork processes.

9. Check the alert log for the database. If check point is taking long time then the issue may be with improper configuration of redo log size and groups

10. check sm58 , if you find many trfcs in executying phase then check is there any issue with the destination system.

11. Check basis update statistics jobs.

12. Check in ST04 if you find few sqls which are using more than 5 % of overall database load then you need to tune the SQL. Check if the SQL stmt is using for a proper index.

13. Check which are the main tables which has been used by the users who are facing performance issue. CHeck there update statistics and indexes storage quality. Also check the fragmentation for tables. If required rebuild index , reoraganize tables , update statistics for the tables.

14. Check the IO . If there is bad response time for few disks then contatct to your IO team

There are many more option to improve the system performance.

Regards

Atul

Answers (3)

Answers (3)

ashish_vikas
Active Contributor
0 Kudos

If i get a complain from end user for poor performance of system.. i try to drill down like this :

Check ST03N : here first i try to find out transaction user is executing and are taking more time(most of the time complete system is not slow ..only a few transaction is slow). After this i check what component of that transaction is high.. DB time or GUI time or wait time..

for example : high db time points towards expansive SQLs and high wait time towards possible unavailability of workprocess.

Also, i know if this is a Z* transaction or standard one.. so now i have to work with SAP note or ABAP person.

many times, for Z* transactions, we get expansive SQL easily with ST04 transaction which can be tuned by ABAPer with creating new Indexes.

Also, few time just rebuilding indexes & stats run for table help.

thanks

ashish

Former Member
0 Kudos

Hello Sisab Nimda .

I think the following T-codes may prove helpful to you,

ST01 - Performance Measurement on the database tables and Buffers through System Trace.

ST04 - Performance Analysis, from where you get largest tables, fastest growing tables, etc.....and accordingly you can act

upon, which will boost performance.

ST14 - Application Analysis, from where you get top cubes, tables, psa tables and many more.

ST10 - Table call statistics, which help you in knowing the communication.

ST30 - Helps in understanding the load on the database.

Best Regards,

Ram

Sriram2009
Active Contributor
0 Kudos

Hi Sisab Nimda

1.You can use the T code STAD type the user name and transaction code which he is using you can check the result how much time taken in the process and database

2. During that time you can check the ST02 for memory tuning / st03 overall work load during the time\

Regards

S