Interpreting Response Times

Response Time and Its Components: Dispatcher Wait Time, Roll-In Time, load Time, Database Time, Processing Time, CPU Time, and Roll-Out Time

Guideline Values for Analyzing Average Response Times for Task Type Dialog

Time	Guideline Value	Problem Indicated
Dispatcher wait time (Av. WAIT TIME)	< 10% of response time; <50 ms	General performance problem with many possible causes
Load time (Av. LOAD & GEN.TIME).	<50 ms	Program buffer too small or CPU bottleneck
Roll -in time, roll-out time (Av. ROLL-IN TIME,Av. ROLL-OUT TIME)	< 20 ms	SAP roll buffer or SAP extended memory too small or CPU bottleneck
Roll wait time (Av. ROLL WAIT TIME)	< 200 ms	Problem with frontend communication (together with higher GU I time) or with communication with external component
GUI time (Av. GU I TIME)	< 200 ms	Problem with frontend communication  (together with higher roll wait time)
Enqueue time  (Av. LOCK TIME)	< 5 ms	Problem with enqueue; network problems
Processing time, CPU time	Processing time < 2 x CPU time	CPU bottleneck or communication problem
Database time AV. DB TIME)	< 40% of response time minusdispatcher wait time; guideline value: 200-600 ms	Database problem,network problem,or CPU bottleneck
Direct read	< 2 ms	Database problem
Sequential read	< 10 ms	Database problem
Logical changes	< 25 ms	Database problem

Working with guideline values

If the values you observe in the workload monitor are significantly above the guideline values, there may be a performance problem in the relevant area (for example, in the database). Note that these values are based on standard situations and may differ in some SAP solutions.

Search for Time Lost

In addition to evaluating the response times, you should perform the following analysis, which could be referred to as the "search for time lost." As previously mentioned, there are two different sources of time statistics. All times, except for CPU time, are measured from the perspective of the SAP work process.CPU t ime is measured from the perspective of the operating system. The lost time analysis checks whether the two statistics can be brought together. To do so, subtract all times or which the SAP work process does not require any CPU time, namely, dispatcher wait time, database time, enqueue time, and roll wait time, from the total average response time. Generally, programs are processed during processing time, and CPU capacity is normally consumed during this time. Thus, processing time and CPU time
should be more or less the same. As a rule of thumb, the difference between processing time and CPU time should not be more than 10%. Greater lost times indicate performance problems. 

What are the possible causes of a significant difference between processing time and CPU time?
->A CPU bottleneck may be a possible cause. In this case, the SAP work processes have insufficient CPU capacity and have to wait for CPU. In t his case, processing time is measured in the work process while no CPU time is used, and this processing time is considerably greater than CPU time.
->Another reason for a difference between processing time and CPU time can be wait times in the SAP work process. Whenever the SAP work process has a STOPPED status, processing time is measured without CPU time being used. This type of wait situation can be identified in the work process overview.