Single-Record Statistics

For every transaction step executed in the SAP system, a record with statistical information is generated and saved in files on the application servers. This statistical information includes response times, memory requirements, database accesses, and so on. By default, there are 48 files, which each contain data for one hour; that is, after 48 hours, the oldest file is overwritten again. Via the stat/maxfiles parameter, you can configure a maximum of 99 statistics files so that the single-record statistics are held for a longer period of time. These records are collected hourly by the collector program RSCOLL00 and deleted after approximately one day.

Displaying statistical records

You can display the statistical records via Transaction STAD, provided they are not yet deleted. After you have called Transaction STAD, you are taken to a selection screen. Restrict the search in this screen, for example, to a user, transaction, or program name, and the period of time that you want to analyze. In the selection mask, you can also establish how the single statistics will be presented. Choose among the options SHOW ALL STATISTIC RECORDS, SORTED BY START TIME (default option), SHOW
ALL RECORDS, GROUPED BY BUSINESS TRANSACTION, and SHOW BUSINESS TRANSACTION SUMM.

A screen with the statistical records that match your selection criteria is then displayed. Click the Sn. FIELDS button to select which statistical values you want displayed in the list. 

Analysis options

With single-record statistics, you can identify problems that might not be visible in the average values of the transaction profile. For example,single records enable you to determine whether the response times for all transaction steps are equally high or generally low but occasionally extremely high (in which case, the averages would be deceptively high). For example, the FCOD column, which displays the function code within a transaction, helps determine whether observed high response times are always associated with a particular transaction screen. 

In the example  the response times for the transaction are well under one second in the first transaction steps, but the last transaction step in the dialog task (marked with function code SICH) displays a high response time of 14.5 seconds. This record should be examined in greater detail.

Statistical Single Records (for Transactions VA01, VA03, etc.)

Detailed view for single record

To this end, double-click an individual statistical record and select the ALL DETAILS button to display all details of the single record

Typical problems

->High database times usually indicate a database problem, which you can analyze using an SQL trace. Using the values for KBYTES TRANSFERRED or DATABASE Rows  you can distinguish between two types of database problems:
--> Database time is high, even though relatively little data is transferred. blew shows a single record for which 14,177 ms are needed for 19 records (in the SEQUENTIAL READ area), which implies an average read time of 746.2 ms per record. 

Single Statistical Record with High Database Time Due t o High Read Time per Data Record (Average Time/Row [ms) = 746.2 for Sequential Read}

-->Database time is high because the quantity of data transferred is large, but the data transfer speed is optimal. An optimal rate of data transfer is a round 1 ms per record. This observation indicates an expensive SQL statement of type 1

->If you find high database times only sporadically, check individual records to see whether the entry NOTE: TABLES WERE SAVED IN THE TABLE BUFFER can be found. This entry means that tables transferred from the database have been saved in the table buffer. If the SAP system has been running for some time with high load since the start, all necessary tables should be located in the table buffer and do not have to be reloaded. If this entry occurs frequently in production operation, a problem with displacements or the invalidation of table buffers is indicated .

->High roll wait times and high GUI times indicate communication problems. You analyze such problems with the RFC trace. 
->High CPU times indicate either time-consuming calculations in the ABAP coding or frequent access to table buffers. Programs with more than 50% CPU time can be examined in greater detail with the ABAP trace or debugger.
->Other problems that can be located using single statistics include those with program buffer load procedures, SAP lock administration (enqueue), or RFC calls.

Main memory requirement

In the TASK AND MEMORY INFORMATION section, you will find information on the main memory requirement of the relevant transaction step.

Fields for Main Memory Requirement in Statistical Single Record

Field	Explanation
TOTAL MEMORY USED	Allocated memory at the end of the transaction step
EXTENDED MEMORY IN USE	Allocated extended memory at t he end of the transaction step
PRIVATE MEMORY IN USE	Allocated heap memory at the end of the transaction step
MAX. EXTENDED MEMORY USED IN TRANSACTION	Maximum memory requirement in the transaction ("high water mark ") (for all steps that belong to the same SAP transaction)
MAX. EXTENDED MEMORY USED IN DIALOG STEP	Maximum memory requirement in t he transaction step ("high water mark")
WORK PROCESS IN PRIV MODE	Information about whet her a work process was in PRIV mode
WORK PROCESS RESTARTED	Information about whether a work process was restarted