If you watch the work process overview for several minutes and repeatedly click the REFRESH button to update the display, you can usually determine whether there is a current performance problem related to work processes in the SAP instance; if there is a problem, you can roughly estimate its cause. The main indication of a performance problem is thatall the work processes of a particular type (such as dialog or update)are occupied.
Long database response times
There is a problem in the database area if the work process overview in the ACTION column shows numerous database-related actions, such as SEQUENTIAL READ, DIRECT READ, UPDATE, COMMIT, and WAITING FOR DB LOCK. In this case, start two additional user sessions, the database process monitor and the database lock monitor (exclusive lock waits), to identify any expensive SQL statements or database locks.
Deactivated update service
If all update work processes (UPDs) are occupied, you have a specific problem. Check whether the update was deactivated. To do so, call Transaction SM13. Determine whether you can find the message UPDATING BY SYSTEM DEACTIVATED. If this is the case, an entry exists in the SAP Syslog (Transaction SM21) showing when, by whom, and for what reason the update was deactivated. As soon as the underlying problem is resolved, for example, a database error, you can reactivate the update work process in Transaction SM13.
Problems with SAP memory management
Problems with SAP memory management are often indicated in the work process overview as follows:
->The ACTION/REASON FOR WAITING field frequently displays ROLL-IN or RoLL-OUT (accompanied by a Type 6 semaphore).
->Many work processes are in PRTV mode (the STATUS field displays»STOPPED and the REASON field displays » PRIV).
Ex: Problem with Used-Up Extended Memory
1.almost all work processes are in the private mode state.
2. both t he SAP extended memory and the roll buffer are completely full.
Increasing the size of the extended memory (parameter em/initial_size_MB) or an analysis of the programs with high memory requirements will likely solve t he problem in this case.
Stopped work processes
If a work process listed in the local work process overview has the status STOPPED, the cause is indicated in the REASON fie ld. To obtain a list of the possible reasons, access SAP Help for this column.
Normally, it is not a problem if some work processes have the status STOPPED for short periods of time. However, if the number of work processes that are stopped for thesame reason exceeds 20%, or if these work processes continue to have the statusSTOPPED for a long time, you should analyze the situation in detail.
A single ineffective or defective
work process often starts a chain reaction that stops other work processes. I A single ineffective or defective work process can be found using the TIME field . You can often assume that the work process with the longest runtime (indicated in the TIME column) caused the problem. If the problem is acute, consider manually terminating the defective work process
Completed work processes
If, in the work process overview, you detect numerous terminated work processes (indicated as COMPLETE in the STATUS field) and find that you cannot restart them, it is likely that there is a problem with the SAP kernel or with logging onto the database. In this case, check the trace file. This can be found in the work process overview under the menu option PROCESS • TRACE • DISPLAY. Save this trace file to a local file. The trace file will be overwritten when the work process is restarted, so save the trace file to a secure location to enable subsequent troubleshooting. Look for SAP Notes referring to the problem at the SAP Service Marketplace, or consult SAP.
Incorrect load distribution
In a distributed system with several computers, you may find a work process bottleneck on at least one computer, while other computers have idle work processes. Check how many users are logged onto each SAP instance. On the workload monitor, you can also check how many dialog steps have been executed on each server. If you discover a very uneven distribution load, you should optimize your logon distribution. Use Transaction SMLG to check whether all the servers are available for logon distribution or whether there are any relevant error messages. Then, proceed with reorganizing the logon distribution.
For an overview of current user distribution, see the logon group distribution via the following path:
TOOLS • ADMINISTRATION • CCMS • CONFIGURATION • LOGON GROUPS
(Transaction SMLG) • GoTo • LOAD DISTRIBUTION
Too few work processes
As you can see from the problems described in the previous sections, there are many reasons why all work processes of a given type might be busy. If you have ruled out all the problems discussed so far yet still have a work process bottleneck, it may be that you have not configured enough work processes. In this case, you should increase the number of work processes. However, before doing so, check whether the computer has sufficient CPU and main memory resources. If the CPU is already being 80% utilized, an increase in the number of work processes will more likely further decrease rather than increase performance.
References to Problems That Can Be Detected Using the Work Process Overview
Problem | Title |
Long database response times | Identifying Expensive SQL Statements; Optimizing SQL Statements Locks Optimizing SQL statements |
SAP memory configuration | Analyzing SAP Memory Configuration Memory Management of SAP NetWeaver ABAP Application Server |
Nonoptimal load distribution, too few work processes | Workload Distribution and Remote Function Calls |